If this had been available in 2010, Redis scripting would have been JavaScript and not Lua. Lua was chosen based on the implementation requirements, not on the language ones... (small, fast, ANSI-C). I appreciate certain ideas in Lua, and people love it, but I was never able to like Lua, because it departs from a more Algol-like syntax and semantics without good reasons, for my taste. This creates friction for newcomers. I love friction when it opens new useful ideas and abstractions that are worth it, if you learn SmallTalk or FORTH and for some time you are lost, it's part of how the languages are different. But I think for Lua this is not true enough: it feels like it departs from what people know without good reasons.
I don't love a good deal of Lua's syntax, but I do think the authors had good reasons for their choices and have generally explained them. Even if you disagree, I think "without good reasons" is overly dismissive.
Personally though, I think the distinctive choices are a boon. You are never confused about what language you are writing because Lua code is so obviously Lua. There is value in this. Once you have written enough Lua, your mind easily switches in and out of Lua mode. Javascript, on the other hand, is filled with poor semantic decisions which for me, cancel out any benefits from syntactic familiarity.
More importantly, Lua has a crucial feature that Javascript lacks: tail call optimization. There are programs that I can easily write in Lua, in spite of its syntactic verbosity, that I cannot write in Javascript because of this limitation. Perhaps this particular JS implementation has tco, but I doubt it reading the release notes.
I have learned as much from Lua as I have Forth (SmallTalk doesn't interest me) and my programming skill has increased significantly since I switched to it as my primary language. Lua is the only lightweight language that I am aware of with TCO. In my programs, I have banned the use of loops. This is a liberation that is not possible in JS or even c, where TCO cannot be relied upon.
In particular, Lua is an exceptional language for writing compilers. Compilers are inherently recursive and thus languages lacking TCO are a poor fit (even if people have been valiantly forcing that square peg through a round hole for all this time).
Having said all that, perhaps as a scripting language for Redis, JS is a better fit. For me though Lua is clearly better than JS on many different dimensions and I don't appreciate the needless denigration of Lua, especially from someone as influential as you.
> For me though Lua is clearly better than JS on many different dimensions and I don't appreciate the needless denigration of Lua, especially from someone as influential as you.
Is it needless? It's useful specifically because he is someone influential, and someone might say "Lua was antirez's choice when making redis, and I trust and respect his engineering, so I'm going to keep Lua as a top contender for use in my project because of that" and him being clear on his choices and reasoning is useful in that respect. In any case where you think he has a responsibility to be careful what he says because of that influence, that can also be used in this case as a reason he should definitely explain his thoughts on it then and now.
Formally JavaScript is specified as having TCO as of ES6, although for unfortunate and painful reasons this is spec fiction - Safari implements it, but Firefox and Chrome do not. Neither did QuickJS last I checked and I don't think this does either.
You’re wrong in the way in which many people are wrong when they hear about a thing called “tail-call optimization”, which is why some people have been trying to get away from the term in favour of “proper tail calls” or something similar, at least as far as R5RS[1]:
> A Scheme implementation is properly tail-recursive if it supports an unbounded number of active tail calls.
The issue here is that, in every language that has a detailed enough specification, there is some provision saying that a program that makes an unbounded number of nested calls at runtime is not legal. Support for proper tail calls means that tail calls (a well-defined subgrammar of the language) do not ever count as nested, which expands the set of legal programs. That’s a language feature, not (merely) a compiler feature.
I still think that the language property (or requirement, or behavior as seen by within the language itself) that we're talking about in this case is "unbounded nested calls" and that the language specs doesn't (shouldn't) assume that such property will be satisfied in a specific way, e.g. switching the call to a branch, as TCO usually means.
I don't think you're wrong per se. This is a "correct" way of thinking of the situation, but it's not the only correct way and it's arguably not the most useful.
A more useful way to understand the situation is that a language's major implementations are more important than the language itself. If the spec of the language says something, but nobody implements it, you can't write code against the spec. And on the flip side, if the major implementations of a language implement a feature that's not in the spec, you can write code that uses that feature.
A minor historical example of this was Python dictionaries. Maybe a decade ago, the Python spec didn't specify that dictionary keys would be retrieved in insertion order, so in theory, implementations of the Python language could do something like:
But the CPython implementation did return all the keys in insertion order, and very few people were using anything other than the CPython implementation, so some codebases started depending on the keys being returned in insertion order without even knowing that they were depending on it. You could say that they weren't writing Python, but that seems a bit pedantic to me.
In any case, Python later standardized that as a feature, so now the ambiguity is solved.
It's all very tricky though, because for example, I wrote some code a decade that used GCC's compare-and-swap extensions, and at least at that time, it didn't compile on Clang. I think you'd have a stronger argument there that I wasn't writing C--not because what I wrote wasn't standard C, but because the code I wrote didn't compile on the most commonly used C compiler. The better approach to communication in this case, I think, is to simply use phrases that communicate what you're doing: instead of saying "C", say "ANSI C", "GCC C", "Portable C", etc.--phrases that communicate what implementations of the language you're supporting. Saying you're writing "C" isn't wrong, it's just not communicating a very important detail: what implementations of the compiler can compile your code. I'm much more interested in effectively communicating what compilers can compile a piece of code than pedantically gatekeeping what's C and what's not.
It wouldn't be the first time the specs have gone too far and beyond their perimeter.
C's "register" variables used to have the same issue, and even "inline" has been downgraded to a mere hint for the compiler (which can ignore it and still be a C compiler).
I'd love to hear more how it is, the state of the library ecosystem, language evolution (wasn't there a new major version recently?), pros/cons, reasons to use it compared to other languages.
About tail-calls, in other languages I've found sometimes a conversion of recursive algorithm to a flat iterative loop with stack/queue to be effective. But it can be a pain, less elegant or intuitive than TCO.
Lua isn't my primary programming language now, but it was for a while. My personal experience on the library ecosystem was:
It's definitely smaller than many languages, and this is something to consider before selecting Lua for a project. But, on the positive side: With some 'other' languages I might find 5 or 10 libraries all doing more or less the same thing, many of them bloated and over-engineered. But with Lua I would often find just one library available, and it would be small and clean enough that I could easily read through its source code and know exactly how it worked.
Another nice thing about Lua when run on LuaJIT: extremely high CPU performance for a scripting language.
In summary: A better choice than it might appear at first, but with trade-offs which need serious consideration.
Does the language give any guarantee that TCO was applied? In other words can it give you an error that the recursion is not of tail call form? Because I imagine a probability of writing a recursion and relying on it being TCO-optimized, where it's not. I would prefer if a language had some form of explicit TCO modifier for a function. Is there any language that has this?
At least in Lua then the rule is simply 'last thing a function dose' this is unambiguous. `return f()` is always a tail call and `return f() + 1` never is.
> More importantly, Lua has a crucial feature that Javascript lacks: tail call optimization. There are programs that I can easily write in Lua, in spite of its syntactic verbosity, that I cannot write in Javascript because of this limitation. Perhaps this particular JS implementation has tco, but I doubt it reading the release notes.
> [...] In my programs, I have banned the use of loops. This is a liberation that is not possible in JS or even c, where TCO cannot be relied upon.
This is not a great language feature, IMO. There are two ways to go here:
1. You can go the Python way, and have no TCO, not ever. Guido van Rossum's reasoning on this is outlined here[1] and here[2], but the high level summary is that TCO makes it impossible to provide acceptably-clear tracebacks.
2. You can go the Chicken Scheme way, and do TCO, and ALSO do CPS conversion, which makes EVERY call into a tail call, without language user having to restructure their code to make sure their recursion happens at the tail.
Either of these approaches has its upsides and downsides, but TCO WITHOUT CPS conversion gives you the worst of both worlds. The only upside is that you can write most of your loops as recursion, but as van Rossum points out, most cases that can be handled with tail recursion, can AND SHOULD be handled with higher-order functions. This is just a much cleaner way to do it in most cases.
And the downsides to TCO without CPS conversion are:
1. Poor tracebacks.
2. Having to restructure your code awkwardly to make recursive calls into tail calls.
3. Easy to make a tail call into not a tail call, resulting in stack overflows.
I'll also add that the main reason recursion is preferable to looping is that it enables all sorts of formal verification. There's some tooling around formal verification for Scheme, but the benefits to eliminating loops are felt most in static, strongly typed languages like Haskell or OCaml. As far as I know Lua has no mature tooling whatsoever that benefits from preferring recursion over looping. It may be that the author of the post I am responding to finds recursion more intuitive than looping, but my experience contains no evidence that recursion is more intuitive than looping: which is more intuitive appears to me to be entirely a function of the programmer's past experience.
In short, treating TCO without CPS conversion as a killer feature seems to me to be a fetishization of functional programming without understanding why functional programming is effective, embracing the madness with none of the method.
EDIT: To point out a weakness to my own argument: there are a bunch of functional programming language implementations that implement TCO without CPS conversion. I'd counter by saying that this is a function of when they were implemented/standardized. Requiring CPS conversion in the Scheme standard would pretty clearly make Scheme an easier to use language, but it would be unreasonable in 2025 to require CPS conversion because so many Scheme implementations don't have it and don't have the resources to implement it.
EDIT 2: I didn't mean for this post to come across as negative on Lua: I love Lua, and in my hobby language interpreter I've been writing, I have spent countless hours implementing ideas I got from Lua. Lua has many strengths--TCO just isn't one of them. When I'm writing Scheme and can't use a higher-order function, I use TCO. When I'm writing Lua and can't use a higher order function, I use loops. And in both languages I'd prefer to use a higher order function.
> it feels like it departs from what people know without good reasons.
Lua was first released in 1993. I think that it's pretty conventional for the time, though yeah it did not follow Algol syntax but Pascal's and Ada's (which were more popular in Brazil at the time than C, which is why that is the case)!
Ruby, which appeared just 2 years later, departs a lot more, arguably without good reasons either? Perl, which is 5 years older and was very popular at the time, is much more "different" than Lua from what we now consider mainstream.
I read this comment, about to snap back with an anecdote how I as a 13 year old was able to learn Lua quite easily, and then I stopped myself because that wasn't productive, then pondered what antirez might think of this comment, and then I realized that antirez wrote it.
I think the older you are the harder Lua is to learn. GP didn't say it made wrong choices, just choices that are gratuitously different from other languages in the Algol family.
I’m tickled that one of my favorite developers is commenting on another of my favorites work. Would be great if Nicolas Cannasse were also in this thread!
It sounds like you're trying to articulate why you don't like Lua, but it seems to just boil down to syntax and semantics unfamiliarity?
I see this argument a lot with Lua. People simply don't like its syntax because we live in a world where C style syntax is more common, and the departure from that seem unnecessary. So going "well actually, in 1992 when Lua was made, C style syntax was more unfamiliar" won't help, because in the current year, C syntax is more familiar.
The first language I learned was Lua, and because of that it seems to have a special place in my heart or something. The reason for this is because in around 2006, the sandbox game "Garry's Mod" was extended with scripting support and chose Lua for seemingly the same reasons as Redis.
The game's author famously didn't like Lua, its unfamiliarity, its syntax, etc. He even modified it to add C style comments and operators. His new sandbox game "s&box" is based on C#, which is the language closest to his heart I think.
The point I'm trying to make is just that Lua is familiar to me and not to you for seemingly no objective reason. Had Garry chosen a different language, I would likely have a different favorite language, and Lua would feel unfamiliar and strange to me.
Lua has been a wild success considering it was born in Brazil, and not some high wealth, network-effected country with all its consequent influential muscle (Ruby? Python? C? Rust? Prolog? Pascal? APL? Ocaml? Show me which one broke out that wasn't "born in the G7"). We should celebrate its plucky success which punches waaay above its adoption weight. It didn't blindly lockstep ALGOL citing "adooooption!!", but didn't indulge in revolution either, and so treads a humble path of cooperative independence of thought.
Come to think of it I don't think I can name a single mainstream language other than Lua that wasn't invented in the G7.
It wouldn't fix the issue of semantics, but "language skins"[1][2] are an underexplored area of programming language development.
People go through all this effort to separate parsing and lexing, but never exploit the ability to just plug in a different lexer that allows for e.g. "{" and "}" tokens instead of "then" and "end", or vice versa.
Not "never exploit"; Reason and BuckleScript are examples of different "language skins" for OCaml.
The problem with "skins" is that they create variety where people strive for uniformity to lower the cognitive load. OTOH transparent switching between skins (about as easy as changing the tab sizes) would alleviate that.
> OTOH transparent switching between skins (about as easy as changing the tab sizes) would alleviate that.
That's one of my hopes for the future of the industry: people will be able to just choose the code style and even syntax family (which you're calling skin) they prefer when editing code, and it will be saved in whatever is the "default" for the language (or even something like the Unison Language: store the AST directly which allows cool stuff like de-duplicating definitions and content-addressable code - an idea I first found out on the amazing talk by Joe Armstrong, "The mess we're in" [1]).
Rust, in particular, would perhaps benefit a lot given how a lot of people hate its syntax... but also Lua for people who just can't stand the Pascal-like syntax and really need their C-like braces to be happy.
That's precisely the point of using tabs for indentation: you don't need to fight over it, because it's a local display preference that does not affect the source code at all, so everyone can just configure whatever they prefer locally without affecting other people.
The idea of "skins" is apparently to push that even further by abstracting the concrete syntax.
Do you mean that files produced with "wide" tabs might have hard newlines embedded more readily in longer lines? Or that maybe people writing with "narrow" tabs might be comfortable writing 6-deep if/else trees that wrap when somebody with their tabs set to wider opens the same file?
I also strongly disliked luas syntax at first but now I feel like the meta tables and what not and pcall and all that stuff is kinda worth it. I like everything about Lua except some of the awkward syntax but I find it so much better then JS, but I haven't been a web dev in over a decade
JavaScript in 2010 was a totally different beast, standartization-wise. Lots of sharp corners and blank spaces were still there.
So, even if an implementation like MicroQuickJS existed in 2010, it's unlikely that too many people would have chosen JS over Lua, given all the shortcomings that JavaScript had at the time.
In 1994 at the second WWW conference we presented "An API to Mosaic". It was TCL embedded inside the (only![1]) browser at the time - Mosaic. The functionality available was substantially similar to what Javascript ended up providing. We used it in our products especially for integrating help and preferences - for example HTML text could be describing color settings, you could click on one, select a colour from the chooser and the page and setting in our products would immediately update. In another demo we were able to print multiple pages of content from the start page, and got a standing ovation! There is an alternate universe where TCL could have become the browser language.
For those not familiar with TCL, the C API is flavoured like main. Callbacks take a list of strings argv style and an argc count. TCL is stringly typed which sounds bad, but the data comes from strings in the HTML and script blocks, and the page HTML is also text, so it fits nicely and the C callbacks are easy to write.
[1] Mosaic Netscape 0.9 was released the week before
The Redis test suite is still written in Tcl: https://news.ycombinator.com/item?id=9963162 (although more recently antirez said somewhere he wished he'd written it in C for speed)
> it feels like it departs from what people know without good reasons.
Lua is a pretty old language. In 1993 the world had not really settled on C style syntax. Compared to Perl or Tcl, Lua's syntax seems rather conventional.
Some design decisions might be a bit unusual, but overall the language feels very consistent and predictable. JS is a mess in comparison.
> because it departs from a more Algol-like syntax
Huh? Lua's syntax is actually very Algol-like since it uses keywords to delimit blocks (e.g. if ... then ... end)
I known for very long time that c (and co) inherited the syntax from algol.
But only after long time I tried to check what Algol actually looked like. To my surprise, Algol does not look anything like C to me.
I would be quite interested in the expanded version of “C has inherited syntax from Algol”
Edit: apparently the inheritance from Algol is a formula: lexical scoping + value returning functions (expression based) - parenthesitis. Only last item is about visual part of the syntax.
That's what matters to me, not how similar Lua is to other languages, but that the language is well-designed in its own system of rules and conventions. It makes sense, every part of it contributes to a harmonious whole. JavaScript on the other hand.
When speaking of Algol or C-style syntax, it makes me imagine a "Common C" syntax, like taking the best, or the least common denominator, of all C-like languages. A minimal subset that fits in your head, instead of what modern C is turning out to be, not to mention C++ or Rust.
I don't write modern C for daily use, so I can't really say. But I've been re-learning and writing C99 more these days, not professionally but personal use - and I appreciate the smallness of the language. Might even say C peaked at C99. I mean, I'd be crazy to say that C-like languages after C99, like Java, PHP, etc., are all misguided for how unnecessarily big and complex they are. It might be that I'm becoming more like a caveman programmer as I get older, I prefer dumb primitive tools.
I think criticizing JavaScript has become a way of signaling "I'm a good programmer." Yes, good programmers ten years ago had valid reasons to criticize it. But today, attacking the efforts of skilled engineers who have improved the language (given the constraints and without breaking half of the web) seems unfair. They’ve achieved a Herculean task compared to the Python dev team, which has broken backward compatibility so many times yet failed to create a consistent language, lacking a single right way to do many things.
There's nothing wrong with 1-based indexing. The only reason it seems wrong to you is because you're familiar with 0-based, not because it's inherently worse.
This indeed is not Algol (or rather C) heritage, but Fortran heritage, not memory offsets but indices in mathematical formulae. This is why R and Julia also have 1-based indexing.
Lots of systems I grew up with were 1-indexed and there's nothing wrong with it. In the context of history, C is the anomaly.
I learned the Wirth languages first (and then later did a lot of programming in MOO, a prototype OO 1-indexed scripting language). Because of that early experience I still slip up and make off by 1 errors occasionally w/ 0 indexed languages.
(Actually both Modula-2 and Ada aren't strictly 1 indexed since you can redefine the indexing range.)
It's fine, I can see the advantages. I just think it's a weird level of blindness to act like 1 indexing is some sort of aberration. It's really not. It's actually quite friendly for new or casual programmers, for one.
Pascal, frankly, allowed to index arrays by any enumerable type; you could use Natural (1-based), or could use 0..whatever. Same with Modula-2; writing it, I freely used 0-based indexing when I wanted to interact with hardware where it made sense, and 1-based indexes when I wanted to implement some math formula.
That example only shows the opposite of what it sounds like you’re saying, although you could be getting at a few different true things. Anyway:
- Every property access in JavaScript is semantically coerced to a string (or a symbol, as of ES6). All property keys are semantically either strings or symbols.
- Property names that are the ToString() of a 31-bit unsigned integer are considered indexes for the purposes of the following two behaviours:
- For arrays, indexes are the elements of the array. They’re the properties that can affect its `length` and are acted on by array methods.
- Indexes are ordered in numeric order before other properties. Other properties are in creation order. (In some even nicher cases, property order is implementation-defined.)
{ let a = {}; a['1'] = 5; a['0'] = 6; Object.keys(a) }
// ['0', '1']
{ let a = {}; a['1'] = 5; a['00'] = 6; Object.keys(a) }
// ['1', '00']
This engine restricts JS in all of the ways I wished I could restrict the language back when I was working on JSC.
You can’t restrict JS that way on the web because of compatibility. But I totally buy that restricting it this way for embedded systems will result in something that sparks joy
I bet MQJS will also be very popular. Quite impressive that bro is going to have two JS engines to brag about in addition to a lot of other very useful things!
Yes, quite! Monsieur Bellard is a legend of computer programming. It would be hard to think of another programmer whose body of public work is more impressive than FB.
Unfortunate that he doesn't seem to write publicly about how he thinks about software. I've never seen him as a guest on any podcast either.
I have long wondered who the "Charlie Gordon" who seems to collaborate with him on everything is. Googling the name brings up a young ballet dancer from England, but I doubt that's the person in question.
Since you’re on the topic, what ever happened to the multi threading stuff you were doing on JSC? Did it stop when you left Apple? Is the code still in JSC or did it get taken out?
If anyone wants to try out MicroQuickJS in a browser here's a simple playground interface for executing a WebAssembly compiled version of it: https://tools.simonwillison.net/microquickjs
I was interested to try Date.now() since this is mentioned as being the only part of the Date implementation that is supported but was surprised to find it always returns 0 for your microquickjs version - your quickjs variant appears to return the current unix time.
Well, as Jeff Atwood famously said [0], "any application that can be written in JavaScript, will eventually be written in JavaScript". I guess that applies to embedded systems too
Fabrice is an absolute legend. Most people would be content with just making QEMU, but this guy makes TinyC and FFmpeg and QuickJS and MicroQuickJS and a bunch of other huge projects.
I am envious that I will never anywhere near his level of productivity.
It's unfortunate that he uploaded this without notable commit history, it would be interesting to see how long it takes a programmer of his caliber to bring up a project like this.
That said, judging by the license file this was based on QuickJS anyway, making it a moot comparison.
> It only supports a subset of Javascript close to ES5 [...]
I have not read the code of the solver, but solving YouTube's JS challenge is so demanding that the team behind yt-dlp ditched their JS emulator written in Python.
I'm not an embedded systems guy (besides using esp32 boards) so this might be a dumb question but does something like this open up the possibility of programming an esp32/arduino board with Javascript, like Micro/Circuit Python?
Sort of related: About ten years ago there was a device called the Tessel by Technical Machine which you programmed with Javascript, npm, the whole nine yards. It was pretty clever - the javascript got transpiled to Lua VM bytecode and ran in the Lua VM on the device (a Cortex M3 I believe). I recently had Claude rewrite their old Node 0.8 CLI tools in Rust because I wasn't inclined to do the javascript archeology needed to get the old tools up and running. Of course then I put the Tessel back in its drawer, but fun nonetheless.
For all the praise he gets here, few seem interested in his methods: writing complete programs, based on robust computer science, with minimal dependencies and tooling.
When I first read the source for his original QuickJS implementation I was amazed to discover he created the entirety of JavaScript in a single xxx thousand line C file (more or less).
That was a sort of defining moment in my personal coding; a lot of my websites and apps are now single file source wherever possible/practical.
I honestly think the single file thing is best reserved for C, given how bad the language support for modularity is.
I've had the inverse experience dealing with a many thousand line "core.php" file way back in the day helping debug an expressionengine site (back in the php 5.2ish days) and it was awful.
Unless you have an editor which can create short links in a hierarchical tree from semantic comments to let you organize your thoughts, digging through thousands of lines of code all in the same scope can be exceptionally painful.
C has no problems splitting programs in N files, to be honest.
The reason FB (and myself, for what it is worth) often write single file large programs (Redis was split after N years of being a single file) is because with enough programming experience you know one very simple thing: complexity is not about how many files you have, but about the internal structure and conceptually separated modules boundaries.
At some point you mainly split for compilation time and to better orient yourself into the file, instead of having to seek a very large mega-file. Pointing the finger to some program that is well written because it's a single file, strlongly correlates to being not a very expert programmer.
I split to enforce encapsulation by defining interfaces in headers based on incomplete structure types. So it helps me with he conceptually separated module boundaries. Super fast compilation is another benefit.
It may not be immediately obvious how to approach modularity since it isn't directly accomplished by explicit language features. But, once you know what you're doing, it's possible to write very large programs with good encapsulation, that span many files, and which nevertheless compile quite rapidly (more or less instantaneously for an incremental build).
I'm not saying other languages don't have better modularity, but to say that C's is bad misses the mark.
Unironically JavaScript is quite good for single file projects (albeit a package.json usually needed)
You can do a huge website entirely in a single file with NodeJS; you can stick re-usable templates in vars and absue multi-line strings (template literals) for all your various content and markup. If you get crafty you can embed clientside code in your 'server.js' too or take it to the next level and use C++ multi-line string literals to wrap all your JS ie- client.js, server.js and package.json in a single .cpp file
Is there any as large as possible single source (or normal with amalgamation version) more or less meaningful project that could be compiled directly with rustc -o executable src.rs? Just to compare build time / memory consumption.
Yes, that's why I've asked about possible rust support of creating such version of normal project. The main issue, I'm unaware of comparably large rust projects without 3rdparty dependencies.
I believe ripgrep has only or mostly dependencies that the main author also controls. It's structured so that ripgrep depends on regex crates by the same author, for example.
I agree: he loves to "roll your own" a lot. Re: minimal dependencies - the codebase has a software FP implementation including printing and parsing, and some home-rolled math routines for trigonometric and other transcendental functions.
Honestly, it's a reminder that, for the time it takes, it's incredibly fun to build from scratch and understand through-and-through your own system.
Although you have to take detours from, say, writing a bytecode VM, to writing FP printing and parsing routines...
Because he choose the hardest path. Difficult problems, no shortcuts, ambitious, taking time to complete. Our environment in general is the opposite of that.
We spend a lot of time doing busy work that's part of the process but doesn't actually move the needle. We write a lot of code that manages abstractions, but doesn't do a lot. All of this busy work feels like progress, but it's avoiding the hard work of actually writing working code.
We underestimate how inefficient working in teams is compared with individuals. We don't value skill and experience and how someone who understands a problem well can be orders of magnitude more productive.
I thought Bellard might be behind even llama.cpp (that would be completely expected for Bellard) but it's actually another great who's done that: Georgi Gerganov: https://github.com/ggerganov
I remember LZEXE from those olden days. When I discovered the author of FFmpeg and QEMU also created LZEXE, I was so impressed. I've been using his software for my entire computing career.
It's similar to the respect I have for the work of Anders Hejlsberg, who created Turbo Pascal, with which I learned to program; and also C# and TypeScript.
Always interesting when people as talented as Bellard manage to (apparently) never write a "full-on" GUI-fronted application, or more specifically, a program that sits between a user with constantly shifting goals and workflows and a "core" that can get the job done.
I would not want to dismiss or diminish by any amount the incredible work he has done. It's just interesting to me that the problems he appears to pick generally take the form of "user sets up the parameters, the program runs to completion".
Fabrice, if you're reading this, please consider replacing Rust instead with your own memory safe language.
The design intent of Rust is a powerful idea, and Rust is the best of its class, but the language itself is under-specified[1] which prevents basic, provably-correct optimizations[0]. At a technical level, Rust could be amended to address these problems, but at a social level, there are now too many people who can block the change, and there's a growing body of backwards compatibility to preserve. This leads reasonable people to give up on Rust and use something else[0], which compounds situations like [2] where projects that need it drop it because it's hard to find people to work on it.
Having written low-level high-performance programs, Fabrice Bellard has the experience to write a memory safe language that allows hardware control. And he has the faculties to assess design changes without tying them up in committee. I covet his attentions in this space.
I think of Rust might trigger a new generation of languages that are developed with the hindsight of rust.
The principle of zero cost abstractions avoids a slow slide of compromising abstraction cost, but I think there could be small cost abstractions that would make for a more pragmatic language. Having Rust to point at to show what performance you could be achieving would aid in avoiding bloating abstractions.
You can call 1000 averaged programmers and see if they can write MicroQuickJS using the same amount of time, or call one averaged programmer and see if he/she can write MicroQuickJS to the same quality in his/her life time. 10X, 100X or 1000X measures the productivity of us mortals, not someone like Fabrice Bellard.
For all the praise he's receiving, I think his web design skills have gone overlooked. bellard.org is fast, responsive and presents information clearly. Actually I think the fancier the website, the shittier the software. Examples: Tarsnap - minimal website, brilliant software. Discord - Whitespacey, animation-heavy abomination of a website. Software: hundreds of MB of JS slop, government wiretap+botnet for degenerates.
Interesting. I wonder if mqjs would make it feasible to massively parallelize JavaScript on the GPU. I want to run thousands of simultaneously JS interpreters, each with an isolated heap and some shared memory. There are some research projects [1, 2] in this direction, but they are fairly experimental.
I'm curious what practical purpose you could have for running a js execution engine in an environment that already contains a (substantially faster) js execution engine? Is it just for the joy of doing it (if so good for you, absolutely nothing wrong with that).
WebAssembly also runs in places other than the web, where there isn't a JavaScript interpreter at hand. It'd be nice to have a fast JavaScript engine that integrates inside the WebAssembly sandbox, and can call and be called by other languages targeting WebAssembly.
That way, programs that embed WebAssembly in order to be scriptable can let people use their choice of languages, including JavaScript.
It allows, for example, to create bindings as I did for raylib graphics library. exaequOS can run any program that can be built to WebAssembly It will soon support WASI p1 and p2. So many programming languages will be possible for creating programs targeting exaequOS
Is there not a way to use the browser native js execution environment for that? You lose a non-trivial amount of performance running js inside quickjs inside of wasm vs the browser native js engine. I wouldn't be surprised if that's 10 or even 20 times slower, and of course requires loading more code into the browser (slower startup, more ram usage). Maybe you don't care about that, but all of that is pretty orthogonal to the environments I an embedded engine like this is intended for.
Maybe with plugins. The WebAssembly way is cross platform. You would be very surprised with the performances of WebAssembly. I have built a Fibonacci test program in Rust that runs faster when built to Wasi than the native target on my MacBook
This is because the execution is very predictable, so the JIT in the runtime can emit optimized code with the knowledge of how the code is going to run. Embedding unpredictable code (like a javascript interpreter) typically has substantially worse performance when executing under a JIT. This is in addition to the fact that Quickjs (despite being pretty good) can't match the performance of sophisticated JIT implementations like V8 or JavaScriptCore
I think this is such an important point. I know all about Bellard's main works. I actually have no idea what he looks like, I've also never seen an interview with him, and I've never read about his specific philosophies when it comes to different software engineering topics. In a world of never-ending bloviations from "influencers" and "thought leaders" it's so awesome to see a real example of true excellence.
The Turing Award is given for breakthroughs in computer science, not for "most productive programmer of all time", and it wouldn't be appropriate for Ballard.
Being an engineer and coding at this stage/level is just remarkable- sadly this trade craft is missing in most (big?) companies as you get promoted away into oblivion.
If there were some form of "developed contributions to computing" award, his name is definitely up there. I think there could be a need for such an award - for people who reliably have created the foundations of modern computing. Otherwise it's almost always things from an academic context, which can be a little too abstract.
Between ffmpeg and qemu, I always think of https://xkcd.com/2347/ when I see Fabrice's work. Especially since ffmpeg provides the backbone of almost all video streaming systems today.
When reading through the projects list of JS restrictions for "stricter" mode, I was expecting to see that it would limit many different JS concepts. But in fact none of the things which are impossible in this subset are things I would do in the course of normal programming anyway. I think all of the JS code I've written over the past few years would work out of the box here.
I was surprised by this one that only showed up lower in the document:
- Date: only Date.now() is supported. [0]
I certainly understand not shipping the js date library especially in an embedded environment both for code-size, and practicality reasons (it's not a great date library), but that would be an issue in many projects (even if you don't use it, libraries yo use almost certainly do.
Good catch. I didn't realize that there was a longer list of restrictions below the section called "Stricter mode", and it seems like a lot of String functions I use are missing too.
Anyone know how this compares to Espruino? The target memory footprint is in the same range, at least. (I know very little about the embedded js space, I just use shellyplugs and have them programmed to talk to BLE lightswitches using some really basic Espruino Javascript.)
They are very similar in terms of ROM footprint (esp: 128K vs mqjs: 100K) and min RAM (esp: 8K vs mqjs: 10K), but spec coverage need to be examined in detail to see the actual difference.
People talk about how productive Fabrice Bellard is, but I don't think anyone appreciates just how productive he is.
Here's the commit history for this project
b700a4d (2025-12-22T1420) - Creates an empty project with an MIT license
295a36b (2025-12-22T1432) - Implements the JavaScript engine, the C API, the REPL, and all documentation
He went from zero to a complete JS implementation in just 12 minutes!
I couldn't do that even if you gave me twice as much time.
Okay, but seriously, this is super cool, and I continue to be amazed by Fabrice. I honestly do think it would be interesting to do an analysis of a day or week of Fabrice's commits to see if there's something about his approach that others can apply besides just being a hardworking genius.
That doesn't mean anything. I quite often start with writing a proof-of-concept, and only initialize the git repository when I'm confident the POC will actually lead to something useful. Common sense says that those files already existed at the time of the first commit.
This would never happen because there's zero incentive to do this.
Browsers are complex because they solve a complex problem: running arbitrary applications in a secure manner across a wide range of platforms. So any "simple" browser you can come up with just won't work in the real world (yes, that means being compatible with websites that normal people use).
I have to disagree, AMP showed that even Google had an internal conflict with the results of WHATWG.. It's naturally quite hard to reach agreements on a subset when many parties will prefer to go backwards to everything but there situations like the first iPhone, ebooks, TV browsing, etc, where normal people buy simpler things and groups that use the simpler subset achieve more in total than those stuck in the complex only format.
(There are even a lot of developers who would inherently drop any feature usage as soon as you can get 10% of users to bring down their stats on caniuse.com to bellow ~90%.)
> that means being compatible with websites that normal people use
No, new adhering websites would emerge and word of mouth would do the rest : normal people would see this fast nerd-web and want rid of their bloated day-to-day monster of a web life.
Just like all those normal people want rid of their bloated day-to-day monster of a web and therefore go and do something like, say, install an ad blocker?
Oh right. 99% of people don't do even that, much less switch their life over to entirely new websites.
In 2025, depending on the study, it is said that 31.5~42.7% of internet users now block ads. Nearly one-third of Americans (32.2%) use ad blockers, with desktop leading at 37%.
I think both wearables and AI assistant could be an incentive on one hand, also towards a more HATEOAS web. However, I guess we haven't really figured out how to replace ad revenue as the primary incentive to make things as complex as possible.
Lots of comments talking about how existing browsers can already do this, but the big benefit that current browsers can't give you is the sheer level of speed and efficiency that a highly restricted "lite web" browser could achieve, especially if the restrictions are made with efficiency in mind.
The embedded use case is obvious, but it'd also be excellent for things like documentation — with such a browser you could probably have a dozen+ doc pages open with resource usage below that of a single regular browser tab. Perfect for things that you have sitting open for long periods of time.
There could be a way:
This HTML-lite spec would be subset of current standard so that if you open this HTML lite page in normal browser it would still work. but HTML-lite browser would only open HTML-lite sites, apart from tech itch it could be used in someplace where not full browser is needed, especially if you are control content generation.
- TV screens UI
- some game engines embed chrome embed thing ( steam store page kind)
- some electron apps / lighter cross platform engine
- less sucky QML
- i think weechat or sth has own xml bashed app froamework thing (so could be useful to people wanting to build everything app app platform
- much richer markdown format ?
WML/WAP got a bad rap I think, largely because of the way it was developed and imposed/introduced.
But it was not insane, and it represented a clarity of thought that then went missing for decades. Several things that were in WML are quite reminiscent of interactions designed in web components today.
Not likely to happen. There is geminiprotocol with gemtext though for those of us that are fine with that level of simplicity.
Work towards an eventual feature freeze and final standardisation of the web would be fantastic though, and a huge benefit to pretty much everyone other than maybe the Chrome developers.
I would actually merge html and js in a single language and bring the layout part of css too (something like having grid and flexbox be elements themselves instead of display styles, more typst kind of showed this is possible in a nice way) and keep css only for the styling part.
Years ago I wrote a tiny xhtml-basic browser for a job. It was great. Some of my best work. But then the iPhone came out and xhtml-basic died practically overnight.
In the earlier days of the web, there were a lot more plugins you'd install to get around on most websites: not just Flash, but things like PDF viewers, Real Video, etc. You'd regularly have to install new codecs, etc. To say nothing of the days when there were some sites you'd have to use a different browser for. A movement towards more of a standards-driven web (in the sense of de facto, not academic, standards) is what made most of this possible.
I think there needs to be a split between the web browser as a document renderer and link follower, and the web browser as a portable target for GUI applications. But frankly my biggest gripe is that you need HTML, JS, and CSS. Three distinct languages that are extremely dissimilar in syntax and semantics that you need all three of (or some bastard cross compiler for your JSX to convert from one format to them). Just make a decent scripting language and interface for the browser and you don't need that nonsense.
I understand this has been tried before (flash, silverlight, etc). They weren't bad ideas, they were killed because of companies that were threatened by the browser as a standard target for applications.
I think this is the ideal direction mainly because a lot of the webs current tech problems stem from websites that don't need app-level features using them. I was in web dev at the advent of SPA-style navigation and understand why everyone switched to it, but at the same time I feel like it's the source of many if not most bugs an performance issues that frustrate the average user.
I mean, you can do all that now, so that's not the problem. The problem would be convincing millions of people to switch, when 99.99999% of them couldn't care less.
My idea is to use Markdown over HTTP(S). It's relatively easy to implement Markdown renderer, compared to HTML renderer. It's possible to browse that kind of website with HTML browser with very simple wrapper either on client or server side, so it's backwards compatible. It's rich enough for a lot of websites with actually useful content.
Now I know that Markdown generally can include HTML tags, so probably it should be somewhat restricted.
It could allow to implement second web in a compatible way with simple browsers.
A few too many 9s there I think. You're estimating that only 1 person in every 10 million could care less. So less than 50 such people in the USA for example
But most "apps" are just webviews running overcomplicated websites in them, many of which are using all the crazy features that the GP post wants to strip out.
And, I don't have to run a binary to try your product. The web has a lot of flaws, but it's a good way to deliver properly sandboxed applications with low hassle on the part of the user. I've built my fair share of native vs web apps, and I vastly prefer working on web apps. As a user, I vastly prefer web apps for most things. Not all things, but most. No, I don't want to install your crappy app on my computer and risk you doing something irresponsible. I'll keep you sandboxed in a browser tab that I can easily "uninstall" by closing.
I will pick a web app over a proprietary "native" app every time. That way, it can stay in a sandbox where it belongs. Discord, Zoom, Meet, Trello, YouTube, and various others, all stay in sandboxed browser tabs.
I have several web apps installed over the native alternatives. Discord is the most prominent one; I've found their native app has been getting shittier by the day over recent months, while the web app remains as snappy as any Safari page. Plus I can run an adblocker and other extensions in the web app which improve the experience.
On a phone at the moment so I can't try it out, but in regards to this "stricter mode" it says global variables must be declared with var. I can't tell if that means that's just the only way to declare a global or if not declaring var makes it scoped in this mode. Based on not finding anything skimming through the examples, I assume the former?
it also talks about the global object not being a place to add properties. So how you might do `window.foo = ...` or `globalThis.foo = ...` to make something from the local context into a global object. in this dialect I guess you would have to reserve any global objects you wanted to set with a `var` and then set them by reference eg
// global. initialized by SomeConstructor
var fooInstance
class SomeConstructor {
constructor(...) {
fooInstance = this;
}
static getInstance(...) {
if (fooInstance != null) return fooInstance;
return new SomeConstructor(...);
}
}
It’s not even the languages or runtimes that inhibit embedded adoption but the software to hardware tooling. Loader scripts, HAL/LL/CMSIS, flashing, etc. They all suck.
It looks like if you write the acceptable MQJS subset of JS+types, then run your code through a checker+stripper that doesn't try to inject implementations of TS's enums and such it should just work?
I don't think you can transpile arbitrary TS in mqjs's JS subset. Maybe you can lint your code in such a way that certain forbidden constructs fail the lint step, but I don't think you can do anything to avoid runtime errors (i.e. writing to an array out of its bonds).
Clarification added later: One of my key interests at the moment is finding ways to run untrusted code from users (or generated by LLMs) in a robust sandbox from a Python application. MicroQuickJS looked like a very strong contender on that front, so I fired up Claude Code to try that out and build some prototypes.
I had Claude Code for web figure out how to run this in a bunch of different ways this morning - I have working prototypes of calling it as a Python FFI library (via ctypes), as a Python compiled module and compiled to WebAssembly and called from Deno and Node.js and Pyodide and Wasmtime https://github.com/simonw/research/blob/main/mquickjs-sandbo...
Down to -4. Is this generic LLM-dislike, or a reaction to perceived over-self-promotion, or something else?
No matter how much you hate LLM stuff I think it's useful to know that there's a working proof of concept of this library compiled to WASM and working as a Python library.
I didn't plan to share this on HN but then MicroQuickJS showed up on the homepage so I figured people might find it useful.
(If I hadn't disclosed I'd used Claude for this I imagine I wouldn't have had any down-votes here.)
Your github research/ links are an interesting case of this. On one hand, late AI adopters may appreciate your example prompts and outputs. But it feels like trivially reproducible noise to expert LLM users, especially if they are unaware of your reputation for substantive work.
The HN AI pushback then drowns out your true message in favor of squashing perceived AI fluff.
This particular case is a very solid use-case for that approach though. There are a ton of important questions to answer: can it run in WebAssembly? What's the difference to regular JavaScript? Is it safe to use as a sandbox against attacks like the regex thing?
Those questions can be answered by having Claude Code crunch along, produce and execute a couple of dozen files of code and report back on the results.
I think the knee-jerk reaction pushing back against this is understandable. I'd encourage people not to miss out on the substance.
Counterpoint to the sibling comment: posting your own site is fine. Your contributions are substantial, and your site is a well-organized repository of your work. Not everything fits (or belongs) in a comment.
I'd chalk up the -4 to generic LLM hate, but I find examples of where LLMs do well to be useful, so I appreciated your post. It displays curiosity, and is especially defensible given your site has no ads, loads blazingly fast, and is filled with HN-relevant content, and doesn't even attempt to sell anything.
And again you're linking to your site. Maybe try pasting the few relevant sentences instead of constantly pushing your content in almost every comment. That's what people find annoying. Maybe link to other people's stuff more, or just write what you think here on HN.
If someone wants to read your blog, they will, they know it exists, and some people even submit your new articles here. There's no need to do what you're doing. Every day you're irritating more people with this behavior, and eventually the substance won't matter to them anymore, so you're acting against your own interests.
Unless you want people to develop the same kind of ad blindness mechanism, where they automatically skip anything that looks like self promotion. Some people will just see a comment by simonw and do the same.
A lot of people have told you this in many threads, but it seems you still don’t get it.
I think you're misreading what the "normalization" problem actually is and why my comment got a lot of upvotes.
You're not pushing against an arbitrary taboo where people dislike self links in principle. People already accept self links on HN when they're occasional and clearly relevant. What people are reacting to is the pattern when "my answer is a link to my site" becomes your default state, it stops reading like helpful reference and starts reading like your distribution strategy.
And that's why "I'm determined to normalize it" probably won't work because you can't normalize your way out of other people's experience of friction. If your behavior reliably adds a speed bump to reading threads forcing people to context switch/click out and wonder if they're being marketed to then the community will develop a shortcut I mentioned in my previous comment which basically is : this is self promo so just ignore.
If your goal is genuinely to share useful ideas, you're better off meeting people where they are: put the relevant 2-6 sentences directly in the comment, and then add something like "I wrote more about it on my blog" or whatever and if anyone is interested they will scroll through your blog (you have it in your profile so anyone can find it with one click) or ask for a link.
Otherwise you're not "normalizing" anything, you're training readers to stop paying attention to you. And I assure you once that happens, it's hard to undo, because people won't relitigate your intent every time. They'll just scroll.
It's a process that's already started, but you can still reverse it.
No, I'm determined to normalize it. I would like a LOT more people to have personal websites where they write at length about things, and then share links to what they have already written where it is relevant to the conversation.
I'm actively pushing back against the "don't promote your site, don't link to it, restate your content in the comments instead" thing.
I am willing to take on risk to my personal reputation and credibility in support of my goal here.
There might be a bit of growth-hacking resistance here, and maybe some StackOverflow culture as well. Neither should be leveled at you, IMHO. I've followed and admired your work since Datasette launched, and I think you're exhibiting remarkably good judgment in how you discuss topics with links to deeper discussion, and it's in keeping with a long tradition of good practices for the web. Thanks for working to normalize the practice.
A lot of HN people got cut by AI in one way or another, so they seem to have personal beefs with AI. I am talking about not only job shortages but also general humbling of the bloated egos.
> I am talking about not only job shortages but also general humbling of the bloated egos.
I'm gonna give you the benefit for the doubt here. Most of us do not dislike genAI because we were fired or "humbled". Most of us dislike it because a) the terrible environmental impacts, b) the terrible economic impacts, and c) the general non-production-readiness of results once you get past common, well-solved problems
Your stated understanding comes off a little bit like "they just don't like it because they're jealous".
Forget about the AI bit. Do you think it's interesting that MicroQuickJS can be used from Python via FFI or as a compiled module, and can also be compiled to WebAssembly and called from Node.js and Deno and from Pyodide running in a browser?
... and that it provides a useful sandbox in that you can robustly limit both the memory and time allowed, including limiting expensive regular expression evaluation?
I included the AI bit because it would have been dishonest not to disclose how I used AI to figure this all out.
It's interesting but I don't think it belongs as a comment under this post. I can use LLMs to create something tangential for each project posted on HN, and so can everyone else. If we all started doing this then the comment section will quickly become useless and not on point.
Offtopic but I went to your website and saw that you created hackernews-mute and recently I was commenting about how one must have created such an extension and ranted about it. So kudos for you to have created it earlier on.
Have a nice day! Awesome stuff, would keep an eye on your blog, Does your blog/website use mataroa by any chance as there are some similarities even if they are both minimalist but overall nice!
Thank you! I don't use Mataroa but I can see the similarities. My current blog setup is a Python script that parses content written in markdown and emits HTML. The CSS is inspired by the other minimal blogs I see here.
Thanks a lot for checking out my blog/project. Have a great day!
but there is signal in what people are inspired to do upon seeing a new project-- why not simply evaluate the interestingness level of these sorts of mashups on their own terms? it actually feels very "hacker"-y to go out and show people possibilities like this. i have no particular comment on how "interesting" the derivative projects are in this case, but i have a feeling if his original post had been framed more like "i think it's super interesting how easy it is to use via FFI on various runtimes X & Y (oh btw in the spirit of full transparency: i used ai to help me. and you can see more details at <link>). especially because i think everyone who peruses HN with some regularity is likely to know of simon's work in at least some capacity, and i am-- speaking only for myself-- essentially always interested in any sort of project he embarks on, especially his llm-assisted experiments and stuff. but hey-- at the end of the day, all of this value judgment is realized as plainly as possible with +1 / -1 (up- and down-vote) and i guess it just is what it is. if number bad, HN no like. shrug.
I agree that there is signal, and that phrasing his original post as you pointed out would have been better.
My issue is that the cost, in terms of time, for these experiments have really gone down with LLMs. Earlier, if someone played around with the posted project, we knew they spent a reasonable amount of time on it, and thus care about the subject. With LLMs, this is not the case.
Usually I watch your stuff very closely (and positively) because you're pushing the edges of how LLMs can be useful for code (and are a lot more honest/forthwright than most enthusiasts about it Going Horribly Wrong and how much work you need to do to keep on top of it.) This one... looks like a crossbar of random things that don't seem like things anyone would actually want to do? Mentioning the sandboxing bit in the first post would have helped a lot, or anything that said why that particular modes are interesting.
I had been in a similar boat and here are some softwares that I recommend or would discuss with you
https://github.com/libriscv/libriscv (I talked with the author of this project, amazing author fwsgonzo is amazing) and they told me that this has the least latency out of any sandbox at only minor consequence of performance that they know of
Btw for sandboxing, kvm itself feels good too and I had discussed it with them in their discord server when they had mentioned that they were working on a minimal kvm server which has since been open sourced (https://github.com/varnish/tinykvm)
Honestly Simon, Deno hosting/the way deno works is another good interesting tidbit for sandboxing. I wish something like deno's sandboxing capabilities came to python perhaps since python fans can appreciate it.
I will try to look more into your github repository too once I get more free.
> Unfortunately it means those languages will be the permanent coding platforms.
not really,
I suspect training volume has a role in debugging a certain class of errors, so there is an advantage to python/ts/sql in those circumstances: if, as an old boss once told me, you code by the bug method :)
The real problems I've had that hint at training data vs logic have been with poorly documented old versions of current languages.
To me, the most amazing capability is not the code they generate but the facility for natural language analysis.
my experience is that agent tools enable polyglot systems because we can now use the right tool for the job, not just the most familiar.
Simon although I find it interesting. And I respect you in this field. I still feel like the reason people call out AI usage or downvote in this case is that in my honest opinion, it would be also more interesting to see people actually write the code and more so (maintain) it as well and create a whole community/github project around microquickjs wasm itself
I read this post of yours https://simonwillison.net/2025/Dec/18/code-proven-to-work/ and although there is a point that can be made that what you are doing isn't a job and I myself create prototypes of code using AI, long term (in my opinion) what really matters are the maintainance and claim (like your article says in a way, that I can pin point a person whose responsible for code to work)
If I find any bug right now, I wouldn't blame it on you but AI and I have varying amount of trust on it
My opinion on the matter is that for prototyping AI can be considered good use but long term it definitely isn't and I am sure that you share a similar viewpoint.
Perhaps you can build a blog post about the nuance of AI? I imagine that a lot of people might share a similar aspect of AI policy where its okay to tinker with it. I am part of the new generation and trust be told I don't think that there becomes much incentives long term unless someone realizes things of not using AI because using AI just feels so lucrative for especially the youngsters.
I am 17 years old and I am going to go into a decent college with (might I add immense competition to begin with) when I have passion about such topics but only to get dissuaded because the benchmark of solving assignments etc. are done by AI and the signal ratio of universities themselves are reducing but they haven't reduced to the point that they are irrelevant but rather that you need to have a university to try to get a job but companies have either freezed hiring which some point out with LLM's
If you ask me, Long term to me it feels like more people might associate themselves with hobbyist computing and even using AI (to be honest sort of like pewdiepie) without being in the industry.
I am not sure what the future holds for me (or for any of us as a matter of fact) but I guess the point I am trying to state is that there is nuance to the discussion from both sides
Your tireless experimenting (and especially documenting) is valuable and I love to see all of it. The avant garde nature of your recent work will draw the occasional flurry of disdain from more jaded types, but I doubt many HN regulars would think you had anything but good intentions! Guess I am basically just saying.. keep it up.
I didn't downvote you. You're one of "the AI guys" to me on HN. The content of your post is fine, too, but, even if it was sketch, I'd've given you the benefit of the doubt.
I think the people interacting with this post are just more likely to appreciate the raw craftsmanship and talent of an individual like Bellard, and coincidentally might be more critical of the machinery that in their perception devalues it. I count myself among them, but didn’t downvote, as I generally think your content is of high quality.
No I mean post it as an HN post and if anybody cares to see it, they'll upvote that and comment in there. That, instead of pigging backing on other posts to get visibility.
I love it, I find the note interesting, educational, and adds to the discussion in context. Guess you're bound to get a few haters when you share your work in public, but I for one appreciate all your posts, comments, articles, open-source projects.
What is the purpose of compiling this to web assembly? What web assembly runtimes are there where there is not already an easily accessible (substantially faster) js execution environment? I know wasmtime exists and is not tied to a js execution engine like basically every other web assembly implementation, but the uses of wasmtime are not restricted from dependencies like v8 or jsc. Usually web assembly is used for providing sandboxing something a js execution environment is already designed to provide, and is only used when the code that requires sandboxing is native code not javascript. It sounds like a good way to waste a lot of performance for some additional sandboxing, but I can't imagine why you would ever design a system that way if you could choose a different (already available and higher performance) sandbox.
I want to build features - both client- and server-side - where users can provide JavaScript code that I then execute safely.
Just having a WebAssembly engine available isn't enough for this - something has to take that user-provided string of JavaScript and execute it within a safe sandbox.
Generally that means you need a JavaScript interpreter that has itself been compiled to WebAssembly. I've experimented with QuickJS itself for that in the past - demo here: https://tools.simonwillison.net/quickjs - but MicroQuickJS may be interesting as a smaller alternative.
If there's a better option than that I'd love to hear about it!
This is generally the purpose of JavaScript execution environments like v8 or jsc (or quickjs although I understand not trusting that as a sandbox to the same degree). They are specifically intended for executing untrusted scripts (eg web browsers). Web assembly’s sandboxing comes from js sandboxing, since it was originally a feature of the same programs for the same reasons. Wrapping one sandbox in another is what I’m surprised by.
Is it though? I have not personally used these libraries, but a cursory google search reveals several options:
- cloudflare/STPyV8: [0] From cloudflare, intended for executing untrusted code.
- Pythonmonkey: [1] Embeds spidermonkey. Not clearly security focused, but sandboxing untrusted code is literally the point of browser js engines.
It's a little less clear how you would do this from node, but the v8 embedding instructions should work https://v8.dev/docs/embed even if nodejs is already a copy of v8.
... whoa, I don't know how I missed it but I hadn't seen STPyV8 before.
I'd seen PyV8 and ruled it out as very unmaintained.
One of my requirements for a sandbox is that it needs to me maintained by a team of professionals who are running a multi-million dollar business on it. Cloudflare certainly count! I wonder what they use STPyV8 for themselves?
PythonMonkey looks promising too: the documentation says "MVP as of September 2024" so clearly it's intended to be stable for production use at this point.
If this had been available in 2010, Redis scripting would have been JavaScript and not Lua. Lua was chosen based on the implementation requirements, not on the language ones... (small, fast, ANSI-C). I appreciate certain ideas in Lua, and people love it, but I was never able to like Lua, because it departs from a more Algol-like syntax and semantics without good reasons, for my taste. This creates friction for newcomers. I love friction when it opens new useful ideas and abstractions that are worth it, if you learn SmallTalk or FORTH and for some time you are lost, it's part of how the languages are different. But I think for Lua this is not true enough: it feels like it departs from what people know without good reasons.
I don't love a good deal of Lua's syntax, but I do think the authors had good reasons for their choices and have generally explained them. Even if you disagree, I think "without good reasons" is overly dismissive.
Personally though, I think the distinctive choices are a boon. You are never confused about what language you are writing because Lua code is so obviously Lua. There is value in this. Once you have written enough Lua, your mind easily switches in and out of Lua mode. Javascript, on the other hand, is filled with poor semantic decisions which for me, cancel out any benefits from syntactic familiarity.
More importantly, Lua has a crucial feature that Javascript lacks: tail call optimization. There are programs that I can easily write in Lua, in spite of its syntactic verbosity, that I cannot write in Javascript because of this limitation. Perhaps this particular JS implementation has tco, but I doubt it reading the release notes.
I have learned as much from Lua as I have Forth (SmallTalk doesn't interest me) and my programming skill has increased significantly since I switched to it as my primary language. Lua is the only lightweight language that I am aware of with TCO. In my programs, I have banned the use of loops. This is a liberation that is not possible in JS or even c, where TCO cannot be relied upon.
In particular, Lua is an exceptional language for writing compilers. Compilers are inherently recursive and thus languages lacking TCO are a poor fit (even if people have been valiantly forcing that square peg through a round hole for all this time).
Having said all that, perhaps as a scripting language for Redis, JS is a better fit. For me though Lua is clearly better than JS on many different dimensions and I don't appreciate the needless denigration of Lua, especially from someone as influential as you.
> For me though Lua is clearly better than JS on many different dimensions and I don't appreciate the needless denigration of Lua, especially from someone as influential as you.
Is it needless? It's useful specifically because he is someone influential, and someone might say "Lua was antirez's choice when making redis, and I trust and respect his engineering, so I'm going to keep Lua as a top contender for use in my project because of that" and him being clear on his choices and reasoning is useful in that respect. In any case where you think he has a responsibility to be careful what he says because of that influence, that can also be used in this case as a reason he should definitely explain his thoughts on it then and now.
Formally JavaScript is specified as having TCO as of ES6, although for unfortunate and painful reasons this is spec fiction - Safari implements it, but Firefox and Chrome do not. Neither did QuickJS last I checked and I don't think this does either.
>Lua is the only lightweight language that I am aware of with TCO.
Scheme is pretty lightweight.
> Lua has a crucial feature that Javascript lacks: tail call optimization.
I'm not familiar with Lua, but I expect tco to be a feature of the compiler, not of the language. Am I wrong?
You’re wrong in the way in which many people are wrong when they hear about a thing called “tail-call optimization”, which is why some people have been trying to get away from the term in favour of “proper tail calls” or something similar, at least as far as R5RS[1]:
> A Scheme implementation is properly tail-recursive if it supports an unbounded number of active tail calls.
The issue here is that, in every language that has a detailed enough specification, there is some provision saying that a program that makes an unbounded number of nested calls at runtime is not legal. Support for proper tail calls means that tail calls (a well-defined subgrammar of the language) do not ever count as nested, which expands the set of legal programs. That’s a language feature, not (merely) a compiler feature.
[1] https://standards.scheme.org/corrected-r5rs/r5rs-Z-H-6.html#...
Thank you for the precise answer.
I still think that the language property (or requirement, or behavior as seen by within the language itself) that we're talking about in this case is "unbounded nested calls" and that the language specs doesn't (shouldn't) assume that such property will be satisfied in a specific way, e.g. switching the call to a branch, as TCO usually means.
I don't think you're wrong per se. This is a "correct" way of thinking of the situation, but it's not the only correct way and it's arguably not the most useful.
A more useful way to understand the situation is that a language's major implementations are more important than the language itself. If the spec of the language says something, but nobody implements it, you can't write code against the spec. And on the flip side, if the major implementations of a language implement a feature that's not in the spec, you can write code that uses that feature.
A minor historical example of this was Python dictionaries. Maybe a decade ago, the Python spec didn't specify that dictionary keys would be retrieved in insertion order, so in theory, implementations of the Python language could do something like:
But the CPython implementation did return all the keys in insertion order, and very few people were using anything other than the CPython implementation, so some codebases started depending on the keys being returned in insertion order without even knowing that they were depending on it. You could say that they weren't writing Python, but that seems a bit pedantic to me.In any case, Python later standardized that as a feature, so now the ambiguity is solved.
It's all very tricky though, because for example, I wrote some code a decade that used GCC's compare-and-swap extensions, and at least at that time, it didn't compile on Clang. I think you'd have a stronger argument there that I wasn't writing C--not because what I wrote wasn't standard C, but because the code I wrote didn't compile on the most commonly used C compiler. The better approach to communication in this case, I think, is to simply use phrases that communicate what you're doing: instead of saying "C", say "ANSI C", "GCC C", "Portable C", etc.--phrases that communicate what implementations of the language you're supporting. Saying you're writing "C" isn't wrong, it's just not communicating a very important detail: what implementations of the compiler can compile your code. I'm much more interested in effectively communicating what compilers can compile a piece of code than pedantically gatekeeping what's C and what's not.
If the language spec requires TCO, I think you can reasonably call it part of the language.
It wouldn't be the first time the specs have gone too far and beyond their perimeter.
C's "register" variables used to have the same issue, and even "inline" has been downgraded to a mere hint for the compiler (which can ignore it and still be a C compiler).
> as my primary language
I'd love to hear more how it is, the state of the library ecosystem, language evolution (wasn't there a new major version recently?), pros/cons, reasons to use it compared to other languages.
About tail-calls, in other languages I've found sometimes a conversion of recursive algorithm to a flat iterative loop with stack/queue to be effective. But it can be a pain, less elegant or intuitive than TCO.
Lua isn't my primary programming language now, but it was for a while. My personal experience on the library ecosystem was:
It's definitely smaller than many languages, and this is something to consider before selecting Lua for a project. But, on the positive side: With some 'other' languages I might find 5 or 10 libraries all doing more or less the same thing, many of them bloated and over-engineered. But with Lua I would often find just one library available, and it would be small and clean enough that I could easily read through its source code and know exactly how it worked.
Another nice thing about Lua when run on LuaJIT: extremely high CPU performance for a scripting language.
In summary: A better choice than it might appear at first, but with trade-offs which need serious consideration.
Re: TCO
Does the language give any guarantee that TCO was applied? In other words can it give you an error that the recursion is not of tail call form? Because I imagine a probability of writing a recursion and relying on it being TCO-optimized, where it's not. I would prefer if a language had some form of explicit TCO modifier for a function. Is there any language that has this?
At least in Lua then the rule is simply 'last thing a function dose' this is unambiguous. `return f()` is always a tail call and `return f() + 1` never is.
Although it’s a bit weird, Able Forth has the explicit word ~
https://github.com/ablevm/able-forth/blob/current/forth.scr
I do prefer this as it keeps the language more regular (fewer surprises)
Sounds a bit like Clojure's "recur". https://clojuredocs.org/clojure.core/recur
Scala has the @tailrec annotation which will raise a warning if the function can’t be TCO’d
> More importantly, Lua has a crucial feature that Javascript lacks: tail call optimization. There are programs that I can easily write in Lua, in spite of its syntactic verbosity, that I cannot write in Javascript because of this limitation. Perhaps this particular JS implementation has tco, but I doubt it reading the release notes.
> [...] In my programs, I have banned the use of loops. This is a liberation that is not possible in JS or even c, where TCO cannot be relied upon.
This is not a great language feature, IMO. There are two ways to go here:
1. You can go the Python way, and have no TCO, not ever. Guido van Rossum's reasoning on this is outlined here[1] and here[2], but the high level summary is that TCO makes it impossible to provide acceptably-clear tracebacks.
2. You can go the Chicken Scheme way, and do TCO, and ALSO do CPS conversion, which makes EVERY call into a tail call, without language user having to restructure their code to make sure their recursion happens at the tail.
Either of these approaches has its upsides and downsides, but TCO WITHOUT CPS conversion gives you the worst of both worlds. The only upside is that you can write most of your loops as recursion, but as van Rossum points out, most cases that can be handled with tail recursion, can AND SHOULD be handled with higher-order functions. This is just a much cleaner way to do it in most cases.
And the downsides to TCO without CPS conversion are:
1. Poor tracebacks.
2. Having to restructure your code awkwardly to make recursive calls into tail calls.
3. Easy to make a tail call into not a tail call, resulting in stack overflows.
I'll also add that the main reason recursion is preferable to looping is that it enables all sorts of formal verification. There's some tooling around formal verification for Scheme, but the benefits to eliminating loops are felt most in static, strongly typed languages like Haskell or OCaml. As far as I know Lua has no mature tooling whatsoever that benefits from preferring recursion over looping. It may be that the author of the post I am responding to finds recursion more intuitive than looping, but my experience contains no evidence that recursion is more intuitive than looping: which is more intuitive appears to me to be entirely a function of the programmer's past experience.
In short, treating TCO without CPS conversion as a killer feature seems to me to be a fetishization of functional programming without understanding why functional programming is effective, embracing the madness with none of the method.
EDIT: To point out a weakness to my own argument: there are a bunch of functional programming language implementations that implement TCO without CPS conversion. I'd counter by saying that this is a function of when they were implemented/standardized. Requiring CPS conversion in the Scheme standard would pretty clearly make Scheme an easier to use language, but it would be unreasonable in 2025 to require CPS conversion because so many Scheme implementations don't have it and don't have the resources to implement it.
EDIT 2: I didn't mean for this post to come across as negative on Lua: I love Lua, and in my hobby language interpreter I've been writing, I have spent countless hours implementing ideas I got from Lua. Lua has many strengths--TCO just isn't one of them. When I'm writing Scheme and can't use a higher-order function, I use TCO. When I'm writing Lua and can't use a higher order function, I use loops. And in both languages I'd prefer to use a higher order function.
[1] https://neopythonic.blogspot.com/2009/04/tail-recursion-elim...
[2] https://neopythonic.blogspot.com/2009/04/final-words-on-tail...
> I do think the authors had good reasons for their choices and have generally explained them
I'm fairly certain antirez is the author of redis
The word "authors" in that phrase refers to the authors of Lua, not Redis.
Pretty sure he's talking about Lua's authors.
> it feels like it departs from what people know without good reasons.
Lua was first released in 1993. I think that it's pretty conventional for the time, though yeah it did not follow Algol syntax but Pascal's and Ada's (which were more popular in Brazil at the time than C, which is why that is the case)!
Ruby, which appeared just 2 years later, departs a lot more, arguably without good reasons either? Perl, which is 5 years older and was very popular at the time, is much more "different" than Lua from what we now consider mainstream.
We had a lot problems embedding Ruby in a multithreaded C program as the garbage collector tries to scan memory between the threads (more details here: https://gitlab.com/nbdkit/nbdkit/-/commit/7364cbaae809b5ffb6... )
Perl, Python, OCaml, Lua and Rust were all fine (Rust wasn't around in 2010 of course).
> Ruby, which appeared just 2 years later, departs a lot more, arguably without good reasons either?
I doubt we ever would have heard about Ruby without it's syntax decisions. From my understanding it's entire raison d'être was readability.
Pascal and Ada are Algol syntaxed relative to most languages.
I read this comment, about to snap back with an anecdote how I as a 13 year old was able to learn Lua quite easily, and then I stopped myself because that wasn't productive, then pondered what antirez might think of this comment, and then I realized that antirez wrote it.
I think the older you are the harder Lua is to learn. GP didn't say it made wrong choices, just choices that are gratuitously different from other languages in the Algol family.
I’m tickled that one of my favorite developers is commenting on another of my favorites work. Would be great if Nicolas Cannasse were also in this thread!
It sounds like you're trying to articulate why you don't like Lua, but it seems to just boil down to syntax and semantics unfamiliarity?
I see this argument a lot with Lua. People simply don't like its syntax because we live in a world where C style syntax is more common, and the departure from that seem unnecessary. So going "well actually, in 1992 when Lua was made, C style syntax was more unfamiliar" won't help, because in the current year, C syntax is more familiar.
The first language I learned was Lua, and because of that it seems to have a special place in my heart or something. The reason for this is because in around 2006, the sandbox game "Garry's Mod" was extended with scripting support and chose Lua for seemingly the same reasons as Redis.
The game's author famously didn't like Lua, its unfamiliarity, its syntax, etc. He even modified it to add C style comments and operators. His new sandbox game "s&box" is based on C#, which is the language closest to his heart I think.
The point I'm trying to make is just that Lua is familiar to me and not to you for seemingly no objective reason. Had Garry chosen a different language, I would likely have a different favorite language, and Lua would feel unfamiliar and strange to me.
GP is the creator of Redis. I would imagine he knows Lua well given that Redis has embedded it for around a decade.
Lua has been a wild success considering it was born in Brazil, and not some high wealth, network-effected country with all its consequent influential muscle (Ruby? Python? C? Rust? Prolog? Pascal? APL? Ocaml? Show me which one broke out that wasn't "born in the G7"). We should celebrate its plucky success which punches waaay above its adoption weight. It didn't blindly lockstep ALGOL citing "adooooption!!", but didn't indulge in revolution either, and so treads a humble path of cooperative independence of thought.
Come to think of it I don't think I can name a single mainstream language other than Lua that wasn't invented in the G7.
It wouldn't fix the issue of semantics, but "language skins"[1][2] are an underexplored area of programming language development.
People go through all this effort to separate parsing and lexing, but never exploit the ability to just plug in a different lexer that allows for e.g. "{" and "}" tokens instead of "then" and "end", or vice versa.
1. <https://hn.algolia.com/?type=comment&prefix=true&query=cxr%2...>
2. <https://old.reddit.com/r/Oberon/comments/1pcmw8n/is_this_sac...>
Not "never exploit"; Reason and BuckleScript are examples of different "language skins" for OCaml.
The problem with "skins" is that they create variety where people strive for uniformity to lower the cognitive load. OTOH transparent switching between skins (about as easy as changing the tab sizes) would alleviate that.
> OTOH transparent switching between skins (about as easy as changing the tab sizes) would alleviate that.
That's one of my hopes for the future of the industry: people will be able to just choose the code style and even syntax family (which you're calling skin) they prefer when editing code, and it will be saved in whatever is the "default" for the language (or even something like the Unison Language: store the AST directly which allows cool stuff like de-duplicating definitions and content-addressable code - an idea I first found out on the amazing talk by Joe Armstrong, "The mess we're in" [1]).
Rust, in particular, would perhaps benefit a lot given how a lot of people hate its syntax... but also Lua for people who just can't stand the Pascal-like syntax and really need their C-like braces to be happy.
[1] https://www.youtube.com/watch?v=lKXe3HUG2l4
Also consider translation to non-English languages, including different writing and syntax systems (e.g. Arabic or Japanese).
Some languages have tools for more or less straightforward skinning.
Clojure to Tamil: https://github.com/echeran/clj-thamil/blob/master/src/clj_th...
C++ to distorted Russian: https://sizeof.livejournal.com/23169.html
People fight about tab sizes all the time though.
That's precisely the point of using tabs for indentation: you don't need to fight over it, because it's a local display preference that does not affect the source code at all, so everyone can just configure whatever they prefer locally without affecting other people.
The idea of "skins" is apparently to push that even further by abstracting the concrete syntax.
> you don't need to fight over it, because it's a local display preference
This has limits.
Files produced with tab=2 and others with tab=8, might have quite different result regarding nesting.
(pain is still on the menu)
Do you mean that files produced with "wide" tabs might have hard newlines embedded more readily in longer lines? Or that maybe people writing with "narrow" tabs might be comfortable writing 6-deep if/else trees that wrap when somebody with their tabs set to wider opens the same file?
One day Brython (python with braces allowing copy paste code to autoindent) will be well supported by LSPs and world peace will ensure
VB.Net is mostly a reskin of C# with a few extras to smooth the transition from VB.
Lowering the barrier to create your own syntax seems like a bad thing though. C.f. perl.
I also strongly disliked luas syntax at first but now I feel like the meta tables and what not and pcall and all that stuff is kinda worth it. I like everything about Lua except some of the awkward syntax but I find it so much better then JS, but I haven't been a web dev in over a decade
The only thing I dislike about Lua is the 1-indexing. I know they had reasons for it but it always caused issues.
JavaScript in 2010 was a totally different beast, standartization-wise. Lots of sharp corners and blank spaces were still there.
So, even if an implementation like MicroQuickJS existed in 2010, it's unlikely that too many people would have chosen JS over Lua, given all the shortcomings that JavaScript had at the time.
Out of interest, was Tcl considered? It's the original embeddable language.
In 1994 at the second WWW conference we presented "An API to Mosaic". It was TCL embedded inside the (only![1]) browser at the time - Mosaic. The functionality available was substantially similar to what Javascript ended up providing. We used it in our products especially for integrating help and preferences - for example HTML text could be describing color settings, you could click on one, select a colour from the chooser and the page and setting in our products would immediately update. In another demo we were able to print multiple pages of content from the start page, and got a standing ovation! There is an alternate universe where TCL could have become the browser language.
For those not familiar with TCL, the C API is flavoured like main. Callbacks take a list of strings argv style and an argc count. TCL is stringly typed which sounds bad, but the data comes from strings in the HTML and script blocks, and the page HTML is also text, so it fits nicely and the C callbacks are easy to write.
[1] Mosaic Netscape 0.9 was released the week before
Wasn't the original Redis prototype written in Tcl?
Yes, previously: https://news.ycombinator.com/item?id=35989909
The Redis test suite is still written in Tcl: https://news.ycombinator.com/item?id=9963162 (although more recently antirez said somewhere he wished he'd written it in C for speed)
> it feels like it departs from what people know without good reasons.
Lua is a pretty old language. In 1993 the world had not really settled on C style syntax. Compared to Perl or Tcl, Lua's syntax seems rather conventional.
Some design decisions might be a bit unusual, but overall the language feels very consistent and predictable. JS is a mess in comparison.
> because it departs from a more Algol-like syntax
Huh? Lua's syntax is actually very Algol-like since it uses keywords to delimit blocks (e.g. if ... then ... end)
I known for very long time that c (and co) inherited the syntax from algol.
But only after long time I tried to check what Algol actually looked like. To my surprise, Algol does not look anything like C to me.
I would be quite interested in the expanded version of “C has inherited syntax from Algol”
Edit: apparently the inheritance from Algol is a formula: lexical scoping + value returning functions (expression based) - parenthesitis. Only last item is about visual part of the syntax.
Algol alternatives were: cobol, fortan, lisp, apl.
The use of curly braces for delimiting blocks of code actually comes from BCPL.
Of course, C also inherited syntax from Algol, but so did most languages.
> consistent and predictable
That's what matters to me, not how similar Lua is to other languages, but that the language is well-designed in its own system of rules and conventions. It makes sense, every part of it contributes to a harmonious whole. JavaScript on the other hand.
When speaking of Algol or C-style syntax, it makes me imagine a "Common C" syntax, like taking the best, or the least common denominator, of all C-like languages. A minimal subset that fits in your head, instead of what modern C is turning out to be, not to mention C++ or Rust.
Is modern C really much more complicated than old C? C++ is a mess of course.
I don't write modern C for daily use, so I can't really say. But I've been re-learning and writing C99 more these days, not professionally but personal use - and I appreciate the smallness of the language. Might even say C peaked at C99. I mean, I'd be crazy to say that C-like languages after C99, like Java, PHP, etc., are all misguided for how unnecessarily big and complex they are. It might be that I'm becoming more like a caveman programmer as I get older, I prefer dumb primitive tools.
Redis' author also made jimtcl, so I don't think the lack of a small engine was the gap
You're replying to Redis' author.
Lua only departs from norms if you’ve had a very narrow experience with other programming languages.
Frankly, I welcome the fact that Redis doesn’t use JavaScript. It’s an abomination of a language. The fewer times I need to use it the better.
I think criticizing JavaScript has become a way of signaling "I'm a good programmer." Yes, good programmers ten years ago had valid reasons to criticize it. But today, attacking the efforts of skilled engineers who have improved the language (given the constraints and without breaking half of the web) seems unfair. They’ve achieved a Herculean task compared to the Python dev team, which has broken backward compatibility so many times yet failed to create a consistent language, lacking a single right way to do many things.
Lua - an entire language off by one.
Not to mention the 1-based indexing sin. JavaScript has a lot of WTFs but they got that right at least.
There's nothing wrong with 1-based indexing. The only reason it seems wrong to you is because you're familiar with 0-based, not because it's inherently worse.
This indeed is not Algol (or rather C) heritage, but Fortran heritage, not memory offsets but indices in mathematical formulae. This is why R and Julia also have 1-based indexing.
Pascal. Modula-2. BASIC. Hell, Logo.
Lately, yes, Julia and R.
Lots of systems I grew up with were 1-indexed and there's nothing wrong with it. In the context of history, C is the anomaly.
I learned the Wirth languages first (and then later did a lot of programming in MOO, a prototype OO 1-indexed scripting language). Because of that early experience I still slip up and make off by 1 errors occasionally w/ 0 indexed languages.
(Actually both Modula-2 and Ada aren't strictly 1 indexed since you can redefine the indexing range.)
It's funny how orthodoxies grow.
In fact zero-based has shown some undeniable advantages over one-based. I couldn't explain it better than Dijkstra's famous essay: http://www.cs.utexas.edu/~EWD/ewd08xx/EWD831.PDF
It's fine, I can see the advantages. I just think it's a weird level of blindness to act like 1 indexing is some sort of aberration. It's really not. It's actually quite friendly for new or casual programmers, for one.
Pascal, frankly, allowed to index arrays by any enumerable type; you could use Natural (1-based), or could use 0..whatever. Same with Modula-2; writing it, I freely used 0-based indexing when I wanted to interact with hardware where it made sense, and 1-based indexes when I wanted to implement some math formula.
And MATLAB. Doesn't make it any better that other languages have the same mistake.
Does it count as 0-indexing when your 0 is a floating point number?
Actually in JS array indexing is same as property indexing right? So it's actually looking up the string '0', as in arr['0']
Huh. I always thought that JS objects supported string and number keys separately, like lua. Nope!
They do, but strings that are numbers will be reinterpreted as numbers.
[edit]
That example only shows the opposite of what it sounds like you’re saying, although you could be getting at a few different true things. Anyway:
- Every property access in JavaScript is semantically coerced to a string (or a symbol, as of ES6). All property keys are semantically either strings or symbols.
- Property names that are the ToString() of a 31-bit unsigned integer are considered indexes for the purposes of the following two behaviours:
- For arrays, indexes are the elements of the array. They’re the properties that can affect its `length` and are acted on by array methods.
- Indexes are ordered in numeric order before other properties. Other properties are in creation order. (In some even nicher cases, property order is implementation-defined.)
If you can't deal with off-by-one errors, you're not a programmer.
Except for Date.
This engine restricts JS in all of the ways I wished I could restrict the language back when I was working on JSC.
You can’t restrict JS that way on the web because of compatibility. But I totally buy that restricting it this way for embedded systems will result in something that sparks joy
He already has a JS engine which doesn’t make these restrictions
Yeah QuickJS is great.
I bet MQJS will also be very popular. Quite impressive that bro is going to have two JS engines to brag about in addition to a lot of other very useful things!
> Quite impressive...
Yes, quite! Monsieur Bellard is a legend of computer programming. It would be hard to think of another programmer whose body of public work is more impressive than FB.
Unfortunate that he doesn't seem to write publicly about how he thinks about software. I've never seen him as a guest on any podcast either.
I have long wondered who the "Charlie Gordon" who seems to collaborate with him on everything is. Googling the name brings up a young ballet dancer from England, but I doubt that's the person in question.
> It would be hard to think of another programmer whose body of public work is more impressive than FB.
Not many, but these do come to mind: Linus Torvalds, Ken Thompson, Dennis Ritchie, Donald Knuth, Rob Pike. But yeah, it’s rarefied air up there.
Since you’re on the topic, what ever happened to the multi threading stuff you were doing on JSC? Did it stop when you left Apple? Is the code still in JSC or did it get taken out?
I never really started on it other than writing up how to do it
If anyone wants to try out MicroQuickJS in a browser here's a simple playground interface for executing a WebAssembly compiled version of it: https://tools.simonwillison.net/microquickjs
It's a variant of my QuickJS playground here: https://tools.simonwillison.net/quickjs
The QuickJS page loads 2.28 MB (675 KB transferred). The MicroQuickJS one loads 303 KB (120 KB transferred).
Looks like those sizes could be improved significantly, as the builds include names etc. I would suggest linking with
emcc -O3
(and maybe even adding --closure 1 )
edit: actually the QuickJS playground looks already optimized - just the MicroQuickJS one could be improved.
Nice. Got it down from 229KB to 148KB! Thanks for the tips.
https://github.com/simonw/research/pull/5
Thats now live on https://tools.simonwillison.net/microquickjs
I was interested to try Date.now() since this is mentioned as being the only part of the Date implementation that is supported but was surprised to find it always returns 0 for your microquickjs version - your quickjs variant appears to return the current unix time.
Well, as Jeff Atwood famously said [0], "any application that can be written in JavaScript, will eventually be written in JavaScript". I guess that applies to embedded systems too
[0] https://en.wikipedia.org/wiki/Jeff_Atwood
Well, wasn't Fabrice Bellard the guy who built a virtual machine with JS so that you could run Linux within the browser?
https://bellard.org/jslinux/vm.html?cpu=riscv64&url=fedora33...
Fabrice is an absolute legend. Most people would be content with just making QEMU, but this guy makes TinyC and FFmpeg and QuickJS and MicroQuickJS and a bunch of other huge projects.
I am envious that I will never anywhere near his level of productivity.
And thanks to that we can run Linux in a PDF as well..
Indeed
https://www.destroyallsoftware.com/talks/the-birth-and-death...
Sounds a bit like rule 35 of the Internet.
It's unfortunate that he uploaded this without notable commit history, it would be interesting to see how long it takes a programmer of his caliber to bring up a project like this.
That said, judging by the license file this was based on QuickJS anyway, making it a moot comparison.
It does say "public repository of..." implying there's a non-public one with real history. Not sure why not upload the main one though.
Maybe he just oneshotted it
Maybe claude code uses bellard as agent
Claude is really Bellard sitting in his kitchen, sipping coffee, casually replying to code requests while getting ready for his day.
This explains a lot. Opus 4.5 gives you access to Bellard after his second cup of coffee.
A sort of reverse code golf where he doesn't care what he sends as all that code will never touch his prod code bases.
I’d expect much better results, honestly
"You're right! I apologize for the confusion. I am, in fact, Fabrice Bellard. Comment allez-vous?"
It's Fabrice so there's a chance he did
I wonder if this could become the most lightweight way for yt-dlp to solve YouTube Javascript challenges.
https://github.com/yt-dlp/yt-dlp/wiki/EJS
(Note that Bellard's QuickJS is already a supported option.)
Not likely:
> It only supports a subset of Javascript close to ES5 [...]
I have not read the code of the solver, but solving YouTube's JS challenge is so demanding that the team behind yt-dlp ditched their JS emulator written in Python.
Likely not, given that it only implements ES5.
I'm not an embedded systems guy (besides using esp32 boards) so this might be a dumb question but does something like this open up the possibility of programming an esp32/arduino board with Javascript, like Micro/Circuit Python?
There are already libraries/frameworks that have supported this:
* espruino (https://www.espruino.com/)
* elk (https://github.com/cesanta/elk)
* DeviceScript (Microsoft Research's now defunct effort, https://github.com/microsoft/devicescript)
That's been possible with Moddable/Kinoma's XS engine, which is standards compliant with ES6 and beyond.
<https://www.moddable.com/faq#comparison>
If you take a look at the MicroQuickJS README, you can see that it's not a full implementation of even ES5, and it's incompatible in several ways.
Just being able to run JS also isn't going to automatically give you any bindings for the environment.
Sort of related: About ten years ago there was a device called the Tessel by Technical Machine which you programmed with Javascript, npm, the whole nine yards. It was pretty clever - the javascript got transpiled to Lua VM bytecode and ran in the Lua VM on the device (a Cortex M3 I believe). I recently had Claude rewrite their old Node 0.8 CLI tools in Rust because I wasn't inclined to do the javascript archeology needed to get the old tools up and running. Of course then I put the Tessel back in its drawer, but fun nonetheless.
Yes. The key enabling feature is a lack of malloc()
Timing really is everything for making the frontpage, I posted this last night and it got no traction.
I suspect it's just random luck, not timing.
Some other guy tried it as well after you, also no luck.
One strategy is to wait for US to wake up, then post, during their morning.
Other strategy is to post the same thing periodically until there is response.
> It compiles and runs Javascript programs with as low as 10 kB of RAM.
Just in time for RAM to become super expensive. How easy would it be to shove this into Chromium and Electron?
Hard because of web compatibility.
The good news is that it would probably not matter much for chromium's memory footprint anyway...
Fabrice Bellard is widely considered one of the most productive and versatile programmers alive:
- FFmpeg: https://bellard.org
- QEMU: https://bellard.org/qemu/
- JSLinux: https://bellard.org/jslinux/
- TCC: https://bellard.org/tcc/
- QuickJS: https://bellard.org/quickjs/
Legendary.
For all the praise he gets here, few seem interested in his methods: writing complete programs, based on robust computer science, with minimal dependencies and tooling.
When I first read the source for his original QuickJS implementation I was amazed to discover he created the entirety of JavaScript in a single xxx thousand line C file (more or less).
That was a sort of defining moment in my personal coding; a lot of my websites and apps are now single file source wherever possible/practical.
I honestly think the single file thing is best reserved for C, given how bad the language support for modularity is.
I've had the inverse experience dealing with a many thousand line "core.php" file way back in the day helping debug an expressionengine site (back in the php 5.2ish days) and it was awful.
Unless you have an editor which can create short links in a hierarchical tree from semantic comments to let you organize your thoughts, digging through thousands of lines of code all in the same scope can be exceptionally painful.
C has no problems splitting programs in N files, to be honest.
The reason FB (and myself, for what it is worth) often write single file large programs (Redis was split after N years of being a single file) is because with enough programming experience you know one very simple thing: complexity is not about how many files you have, but about the internal structure and conceptually separated modules boundaries.
At some point you mainly split for compilation time and to better orient yourself into the file, instead of having to seek a very large mega-file. Pointing the finger to some program that is well written because it's a single file, strlongly correlates to being not a very expert programmer.
I split to enforce encapsulation by defining interfaces in headers based on incomplete structure types. So it helps me with he conceptually separated module boundaries. Super fast compilation is another benefit.
> complexity is not about how many files you have, but about the internal structure and conceptually separated modules boundaries.
You probably don't need this, but ...
https://www.lelanthran.com/chap13/content.html
C's support for modularity is actually rather strong. This PDF gives a good overview of the basic techniques available: http://www.metamodulaire.org/Computing/modular-c.pdf
It may not be immediately obvious how to approach modularity since it isn't directly accomplished by explicit language features. But, once you know what you're doing, it's possible to write very large programs with good encapsulation, that span many files, and which nevertheless compile quite rapidly (more or less instantaneously for an incremental build).
I'm not saying other languages don't have better modularity, but to say that C's is bad misses the mark.
Unironically JavaScript is quite good for single file projects (albeit a package.json usually needed)
You can do a huge website entirely in a single file with NodeJS; you can stick re-usable templates in vars and absue multi-line strings (template literals) for all your various content and markup. If you get crafty you can embed clientside code in your 'server.js' too or take it to the next level and use C++ multi-line string literals to wrap all your JS ie- client.js, server.js and package.json in a single .cpp file
> I honestly think the single file thing is best reserved for C, given how bad the language support for modularity is.
You don't program much in C, do you?
SQLite 3.51.1
Is there any as large as possible single source (or normal with amalgamation version) more or less meaningful project that could be compiled directly with rustc -o executable src.rs? Just to compare build time / memory consumption.The sqlite3.c file is generated from more finely-grained real sources, see https://www.sqlite.org/src/doc/trunk/README.md
SQLite is only deployed as a single file but the original sources are multiple files. They call it "The Amalgamation".
https://sqlite.org/src/doc/trunk/README.md
Yes, that's why I've asked about possible rust support of creating such version of normal project. The main issue, I'm unaware of comparably large rust projects without 3rdparty dependencies.
From my daily-use utilities, ripgrep and bat seem to have zero dependencies.
I believe ripgrep has only or mostly dependencies that the main author also controls. It's structured so that ripgrep depends on regex crates by the same author, for example.
Looking at Cargo.toml, ripgrep seems to have some dependencies and bat has a lot.
This is like Feynman's method for solving hard scientific problems: write down the question, think very hard, write down the answer.
It doesn't necessarily translate to people who are less brilliant.
Yeah, "Step 1: draw 2 circles. Step 2: draw the rest of the fucking owl"
I agree: he loves to "roll your own" a lot. Re: minimal dependencies - the codebase has a software FP implementation including printing and parsing, and some home-rolled math routines for trigonometric and other transcendental functions.
Honestly, it's a reminder that, for the time it takes, it's incredibly fun to build from scratch and understand through-and-through your own system.
Although you have to take detours from, say, writing a bytecode VM, to writing FP printing and parsing routines...
Because he choose the hardest path. Difficult problems, no shortcuts, ambitious, taking time to complete. Our environment in general is the opposite of that.
We spend a lot of time doing busy work that's part of the process but doesn't actually move the needle. We write a lot of code that manages abstractions, but doesn't do a lot. All of this busy work feels like progress, but it's avoiding the hard work of actually writing working code.
We underestimate how inefficient working in teams is compared with individuals. We don't value skill and experience and how someone who understands a problem well can be orders of magnitude more productive.
He's also built a closed-source LLM inference engine, which he's been maintaining since the GPT-2 days: https://bellard.org/ts_server/ and https://textsynth.com/
I used to play around with Textsynth, but not being OSS killed the appeal for me once llama.cpp came around.
I thought Bellard might be behind even llama.cpp (that would be completely expected for Bellard) but it's actually another great who's done that: Georgi Gerganov: https://github.com/ggerganov
He's also won the International Obfuscated C Code Contest 3 times.
https://www.ioccc.org/authors.html#Fabrice_Bellard
Don't forget his LZEXE from the good old DOS days which was an excellent piece of work at the time.
I remember LZEXE from those olden days. When I discovered the author of FFmpeg and QEMU also created LZEXE, I was so impressed. I've been using his software for my entire computing career.
It's similar to the respect I have for the work of Anders Hejlsberg, who created Turbo Pascal, with which I learned to program; and also C# and TypeScript.
Self-decompressing executables felt like magic to me at the time. Fantastic work, overall.
Wikipedia doesn't list any honours awarded by the French Government. Nor do I see anything from ACM. Definitely overdue some official recognition.
Always interesting when people as talented as Bellard manage to (apparently) never write a "full-on" GUI-fronted application, or more specifically, a program that sits between a user with constantly shifting goals and workflows and a "core" that can get the job done.
I would not want to dismiss or diminish by any amount the incredible work he has done. It's just interesting to me that the problems he appears to pick generally take the form of "user sets up the parameters, the program runs to completion".
Don't forget his LLM based text compression software that won awards.
Guy is a genius. I hope he tries Rust someday
Fabrice, if you're reading this, please consider replacing Rust instead with your own memory safe language.
The design intent of Rust is a powerful idea, and Rust is the best of its class, but the language itself is under-specified[1] which prevents basic, provably-correct optimizations[0]. At a technical level, Rust could be amended to address these problems, but at a social level, there are now too many people who can block the change, and there's a growing body of backwards compatibility to preserve. This leads reasonable people to give up on Rust and use something else[0], which compounds situations like [2] where projects that need it drop it because it's hard to find people to work on it.
Having written low-level high-performance programs, Fabrice Bellard has the experience to write a memory safe language that allows hardware control. And he has the faculties to assess design changes without tying them up in committee. I covet his attentions in this space.
[0]: https://databento.com/blog/why-we-didnt-rewrite-our-feed-han...
[1]: https://blog.polybdenum.com/2024/06/07/the-inconceivable-typ...
[2]: https://daniel.haxx.se/blog/2024/12/21/dropping-hyper/
I think of Rust might trigger a new generation of languages that are developed with the hindsight of rust.
The principle of zero cost abstractions avoids a slow slide of compromising abstraction cost, but I think there could be small cost abstractions that would make for a more pragmatic language. Having Rust to point at to show what performance you could be achieving would aid in avoiding bloating abstractions.
> At a technical level, Rust could be amended to address these problems
I don’t think it can, no.
peak hacker news comment lol
I've only been here for a month. I guess I'm learning well
Whenever someone says there's no such thing as a 10x programmer, I point them to Fabrice and they usually change their mind.
Perhaps closer to 100x actually.
You can call 1000 averaged programmers and see if they can write MicroQuickJS using the same amount of time, or call one averaged programmer and see if he/she can write MicroQuickJS to the same quality in his/her life time. 10X, 100X or 1000X measures the productivity of us mortals, not someone like Fabrice Bellard.
If you're in a room with 100 physicists and Feynman, the accumulated wisdom of Feynman is your best bet.
At this point I'm convinced that they're not a 'real person' and the 'Fabrice' is an operational code name for a very mature hacker collective.
Real people have to sleep at some point!
Funny how people know Fabrice for all the software stuff but none of the hardware antics:
played with implementing analog modem DSP in software in 1999 (linmodem is ~50-80% there, sadly never finished)
probably leading to
played with implementing SDR (again DSP) using VGA output to transmit DVB-T/NTSC/PAL in 2005
probably leading to
Amarisoft SDR 5G base station, commercial product started in 2012 - his current job https://www.amarisoft.com/company/about-us
I was going to mention the last one, and his ASN.1 compiler which is likely related to his telco work:
https://bellard.org/ffasn1/
For real. The GOAT is at it again!
The first two links are broken.
The ffmpeg link was changed apparently, but the QEmu link still works he just redirects to the QEmu homepage.
For all the praise he's receiving, I think his web design skills have gone overlooked. bellard.org is fast, responsive and presents information clearly. Actually I think the fancier the website, the shittier the software. Examples: Tarsnap - minimal website, brilliant software. Discord - Whitespacey, animation-heavy abomination of a website. Software: hundreds of MB of JS slop, government wiretap+botnet for degenerates.
The math checks out.
Interesting. I wonder if mqjs would make it feasible to massively parallelize JavaScript on the GPU. I want to run thousands of simultaneously JS interpreters, each with an isolated heap and some shared memory. There are some research projects [1, 2] in this direction, but they are fairly experimental.
[1]: https://github.com/SamGinzburg/VectorVisor
[2]: https://github.com/beehive-lab/ProtonVM
I easily managed to build quickJS to WebAssembly for running in https://exaequOS.com . So I need to do the same for MicroQuickJS !
I'm curious what practical purpose you could have for running a js execution engine in an environment that already contains a (substantially faster) js execution engine? Is it just for the joy of doing it (if so good for you, absolutely nothing wrong with that).
WebAssembly also runs in places other than the web, where there isn't a JavaScript interpreter at hand. It'd be nice to have a fast JavaScript engine that integrates inside the WebAssembly sandbox, and can call and be called by other languages targeting WebAssembly.
That way, programs that embed WebAssembly in order to be scriptable can let people use their choice of languages, including JavaScript.
It allows, for example, to create bindings as I did for raylib graphics library. exaequOS can run any program that can be built to WebAssembly It will soon support WASI p1 and p2. So many programming languages will be possible for creating programs targeting exaequOS
Is there not a way to use the browser native js execution environment for that? You lose a non-trivial amount of performance running js inside quickjs inside of wasm vs the browser native js engine. I wouldn't be surprised if that's 10 or even 20 times slower, and of course requires loading more code into the browser (slower startup, more ram usage). Maybe you don't care about that, but all of that is pretty orthogonal to the environments I an embedded engine like this is intended for.
Maybe with plugins. The WebAssembly way is cross platform. You would be very surprised with the performances of WebAssembly. I have built a Fibonacci test program in Rust that runs faster when built to Wasi than the native target on my MacBook
This is because the execution is very predictable, so the JIT in the runtime can emit optimized code with the knowledge of how the code is going to run. Embedding unpredictable code (like a javascript interpreter) typically has substantially worse performance when executing under a JIT. This is in addition to the fact that Quickjs (despite being pretty good) can't match the performance of sophisticated JIT implementations like V8 or JavaScriptCore
If there were a software engineering hall of fame, I nominate Fabrice.
rare occasion where he gained a legendary status based purely on his work, I dont think I ever saw even a written interview with the guy
He is a private man that does not like the spotlight IIUC. He refuses most requests for interviews, but they do exist.
https://www.macplus.net/depeche-82364-interview-le-createur-...
https://www.mo4tech.com/fabrice-bellard-one-man-is-worth-a-t... (few quotes, more like a profile piece)
He keeps a low profile and let his work speak for itself.
He really is brilliant.
He has probably has no time for interviews and just focuses on working on his many projects.
I think this is such an important point. I know all about Bellard's main works. I actually have no idea what he looks like, I've also never seen an interview with him, and I've never read about his specific philosophies when it comes to different software engineering topics. In a world of never-ending bloviations from "influencers" and "thought leaders" it's so awesome to see a real example of true excellence.
Bellard it the most genius programmer to ever exist, and the least known compared to other pseudo stars.
There is! ACM grants several awards for scientists and more.
One such award is the Turing Award [1], given "for contributions of lasting and major technical importance to computer science."
[1] https://en.wikipedia.org/wiki/Turing_Award
Possibly more relevant is the "ACM Software System Award": https://en.wikipedia.org/w/index.php?title=ACM_Software_Syst...
Linux and Torvalds hasn't gotten one?
The Turing Award is given for breakthroughs in computer science, not for "most productive programmer of all time", and it wouldn't be appropriate for Ballard.
Fabrice is certainly very skilled in CS, but his achievements are more in software implementation IMO.
AIUI the Turing award is primarily CS focused.
His consistency and craftsmanship is amazing.
Being an engineer and coding at this stage/level is just remarkable- sadly this trade craft is missing in most (big?) companies as you get promoted away into oblivion.
If there were some form of "developed contributions to computing" award, his name is definitely up there. I think there could be a need for such an award - for people who reliably have created the foundations of modern computing. Otherwise it's almost always things from an academic context, which can be a little too abstract.
Between ffmpeg and qemu, I always think of https://xkcd.com/2347/ when I see Fabrice's work. Especially since ffmpeg provides the backbone of almost all video streaming systems today.
Except that ffmpeg and qemu are not maintained by Fabrice. He's one of the greatest programmers but he's not maintaining the internet.
I suppose that if he were to maintain any of these projects, we would never see the new frontiers he has been conquering.
When reading through the projects list of JS restrictions for "stricter" mode, I was expecting to see that it would limit many different JS concepts. But in fact none of the things which are impossible in this subset are things I would do in the course of normal programming anyway. I think all of the JS code I've written over the past few years would work out of the box here.
I was surprised by this one that only showed up lower in the document:
- Date: only Date.now() is supported. [0]
I certainly understand not shipping the js date library especially in an embedded environment both for code-size, and practicality reasons (it's not a great date library), but that would be an issue in many projects (even if you don't use it, libraries yo use almost certainly do.
https://github.com/bellard/mquickjs/blob/main/README.md#:~:t...
Good catch. I didn't realize that there was a longer list of restrictions below the section called "Stricter mode", and it seems like a lot of String functions I use are missing too.
Anyone know how this compares to Espruino? The target memory footprint is in the same range, at least. (I know very little about the embedded js space, I just use shellyplugs and have them programmed to talk to BLE lightswitches using some really basic Espruino Javascript.)
What is the difference with https://www.espruino.com/ ?
They are very similar in terms of ROM footprint (esp: 128K vs mqjs: 100K) and min RAM (esp: 8K vs mqjs: 10K), but spec coverage need to be examined in detail to see the actual difference.
I wonder if AI is so good why dont we have this kind of software released by people here who yap about it.
People talk about how productive Fabrice Bellard is, but I don't think anyone appreciates just how productive he is.
Here's the commit history for this project
b700a4d (2025-12-22T1420) - Creates an empty project with an MIT license
295a36b (2025-12-22T1432) - Implements the JavaScript engine, the C API, the REPL, and all documentation
He went from zero to a complete JS implementation in just 12 minutes!
I couldn't do that even if you gave me twice as much time.
Okay, but seriously, this is super cool, and I continue to be amazed by Fabrice. I honestly do think it would be interesting to do an analysis of a day or week of Fabrice's commits to see if there's something about his approach that others can apply besides just being a hardworking genius.
It's funny how many people replying here just got whooshed. This comment is satire; they don't actually think Bellard wrote everything in 12 minutes.
Doesn't say much. Probably had it largely written down and put it together. I don't think it'd even be humanely possible to do that in 12 minutes.
That doesn't mean anything. I quite often start with writing a proof-of-concept, and only initialize the git repository when I'm confident the POC will actually lead to something useful. Common sense says that those files already existed at the time of the first commit.
When you code with butterflies, you code at the speed of flight.
https://www.xkcd.com/378/
I wonder how do you not understand this was a joke. Did you not fully read it? Or it's just not obvious enough?
I wish for this new year we reboot the Web with a super light standard and accompanying ecosystem with
That would make my year.This would never happen because there's zero incentive to do this.
Browsers are complex because they solve a complex problem: running arbitrary applications in a secure manner across a wide range of platforms. So any "simple" browser you can come up with just won't work in the real world (yes, that means being compatible with websites that normal people use).
I have to disagree, AMP showed that even Google had an internal conflict with the results of WHATWG.. It's naturally quite hard to reach agreements on a subset when many parties will prefer to go backwards to everything but there situations like the first iPhone, ebooks, TV browsing, etc, where normal people buy simpler things and groups that use the simpler subset achieve more in total than those stuck in the complex only format.
(There are even a lot of developers who would inherently drop any feature usage as soon as you can get 10% of users to bring down their stats on caniuse.com to bellow ~90%.)
> that means being compatible with websites that normal people use
No, new adhering websites would emerge and word of mouth would do the rest : normal people would see this fast nerd-web and want rid of their bloated day-to-day monster of a web life.
One can still hope..
Just like all those normal people want rid of their bloated day-to-day monster of a web and therefore go and do something like, say, install an ad blocker?
Oh right. 99% of people don't do even that, much less switch their life over to entirely new websites.
> 99% of people
In 2025, depending on the study, it is said that 31.5~42.7% of internet users now block ads. Nearly one-third of Americans (32.2%) use ad blockers, with desktop leading at 37%.
Wow. That's way higher than I thought. Huh!
It actually gives me hope that we may find a way out of the enshittification of the web.
I don't care to run an ad blocker because sites are still bloated and slow.
I think both wearables and AI assistant could be an incentive on one hand, also towards a more HATEOAS web. However, I guess we haven't really figured out how to replace ad revenue as the primary incentive to make things as complex as possible.
Zero incentive seems a little strong,
Lots of comments talking about how existing browsers can already do this, but the big benefit that current browsers can't give you is the sheer level of speed and efficiency that a highly restricted "lite web" browser could achieve, especially if the restrictions are made with efficiency in mind.
The embedded use case is obvious, but it'd also be excellent for things like documentation — with such a browser you could probably have a dozen+ doc pages open with resource usage below that of a single regular browser tab. Perfect for things that you have sitting open for long periods of time.
That's it. Plus they would work neatly on old computers/phones.
someone should embed it into dillo!
There could be a way: This HTML-lite spec would be subset of current standard so that if you open this HTML lite page in normal browser it would still work. but HTML-lite browser would only open HTML-lite sites, apart from tech itch it could be used in someplace where not full browser is needed, especially if you are control content generation. - TV screens UI - some game engines embed chrome embed thing ( steam store page kind) - some electron apps / lighter cross platform engine - less sucky QML - i think weechat or sth has own xml bashed app froamework thing (so could be useful to people wanting to build everything app app platform - much richer markdown format ?
It’s called WML/WAP
I think we can do better than a 15x15 text window
WML/WAP got a bad rap I think, largely because of the way it was developed and imposed/introduced.
But it was not insane, and it represented a clarity of thought that then went missing for decades. Several things that were in WML are quite reminiscent of interactions designed in web components today.
While we're wishing, can we split CSS into two parts - styling and layout? Also, I'd like to fix the spelling on the "referer" header...
What about https://geminiprotocol.net/
And also bring back progressive enhancement.
https://en.wikipedia.org/wiki/Progressive_enhancement
So you want 2026 to be the year of Google AMP?
Not likely to happen. There is geminiprotocol with gemtext though for those of us that are fine with that level of simplicity.
Work towards an eventual feature freeze and final standardisation of the web would be fantastic though, and a huge benefit to pretty much everyone other than maybe the Chrome developers.
I would actually merge html and js in a single language and bring the layout part of css too (something like having grid and flexbox be elements themselves instead of display styles, more typst kind of showed this is possible in a nice way) and keep css only for the styling part.
Or maybe just make it all a single lispy language
Do it, man. Call it "MicroWeb" or whatever. Write an agent, make it "viewable with regular browsers". I think this could be cool.
Years ago I wrote a tiny xhtml-basic browser for a job. It was great. Some of my best work. But then the iPhone came out and xhtml-basic died practically overnight.
I can't think of an instance of the web contracting like that. Maybe when Apple decided not to support Adobe Flash.
In the earlier days of the web, there were a lot more plugins you'd install to get around on most websites: not just Flash, but things like PDF viewers, Real Video, etc. You'd regularly have to install new codecs, etc. To say nothing of the days when there were some sites you'd have to use a different browser for. A movement towards more of a standards-driven web (in the sense of de facto, not academic, standards) is what made most of this possible.
Arguably XSLT
Would be cool to create a MicroBrowser, just to browser stuff that's compatible.
And Microsoftware running on the Micronet.
You mean like the piece of crap that was WAP?
I think there needs to be a split between the web browser as a document renderer and link follower, and the web browser as a portable target for GUI applications. But frankly my biggest gripe is that you need HTML, JS, and CSS. Three distinct languages that are extremely dissimilar in syntax and semantics that you need all three of (or some bastard cross compiler for your JSX to convert from one format to them). Just make a decent scripting language and interface for the browser and you don't need that nonsense.
I understand this has been tried before (flash, silverlight, etc). They weren't bad ideas, they were killed because of companies that were threatened by the browser as a standard target for applications.
I think this is the ideal direction mainly because a lot of the webs current tech problems stem from websites that don't need app-level features using them. I was in web dev at the advent of SPA-style navigation and understand why everyone switched to it, but at the same time I feel like it's the source of many if not most bugs an performance issues that frustrate the average user.
I agree. Something componenty like Flash, yes. But it'd be easier to subset what already exists..
You can already create websites to these standards. Then truncate large parts of webkit and create a new browser. Or base it on Servo.
I mean, you can do all that now, so that's not the problem. The problem would be convincing millions of people to switch, when 99.99999% of them couldn't care less.
My idea is to use Markdown over HTTP(S). It's relatively easy to implement Markdown renderer, compared to HTML renderer. It's possible to browse that kind of website with HTML browser with very simple wrapper either on client or server side, so it's backwards compatible. It's rich enough for a lot of websites with actually useful content.
Now I know that Markdown generally can include HTML tags, so probably it should be somewhat restricted.
It could allow to implement second web in a compatible way with simple browsers.
You can just use HTML4 if you want, it's already supported and standardized. Markdown is very much not.
I believe this is the way we might get out of this mess.
With a markdown over HTTP browser I could already almost browse Github through the READMEs and probably other websites.
Markdown is really a loved and now quite popular format. It is sad gemini created a separate closed format instead of just adopting it.
A few too many 9s there I think. You're estimating that only 1 person in every 10 million could care less. So less than 50 such people in the USA for example
Maybe you dont need a big enough % to change but a sufficient absolute number, which given internet size might happen with the right 0.00001%
Oh they would care if one shows them much snappier versions of services they use. They just don't know better.
And if you find you need more features than that - just build an app, don't make the web browser into some overly bloated app!
But most "apps" are just webviews running overcomplicated websites in them, many of which are using all the crazy features that the GP post wants to strip out.
Then you have to deal with os compatibility. That's the main selling point of the Web, it works everywhere.
And, I don't have to run a binary to try your product. The web has a lot of flaws, but it's a good way to deliver properly sandboxed applications with low hassle on the part of the user. I've built my fair share of native vs web apps, and I vastly prefer working on web apps. As a user, I vastly prefer web apps for most things. Not all things, but most. No, I don't want to install your crappy app on my computer and risk you doing something irresponsible. I'll keep you sandboxed in a browser tab that I can easily "uninstall" by closing.
I can't think of a single thing where I prefer a web app over a native alternative, unless it's for one-off use.
I will pick a web app over a proprietary "native" app every time. That way, it can stay in a sandbox where it belongs. Discord, Zoom, Meet, Trello, YouTube, and various others, all stay in sandboxed browser tabs.
I have several web apps installed over the native alternatives. Discord is the most prominent one; I've found their native app has been getting shittier by the day over recent months, while the web app remains as snappy as any Safari page. Plus I can run an adblocker and other extensions in the web app which improve the experience.
Most of the “apps” are 200 MB native monstrosities that could be served by 20 kb of JS.
Except when it doesn't because of browser or platform differences/incompatibilities.
The portability of the Web is imperfect, but it's not even in the same galaxy as the portability of native app platforms; there's just no comparison.
Love it! Needing only 10K of RAM, it looks like a much better solution to CircuitPython (can squeeze into 32K).
This is a hell of a Christmas present, Mr. Santa Bellard!
On a phone at the moment so I can't try it out, but in regards to this "stricter mode" it says global variables must be declared with var. I can't tell if that means that's just the only way to declare a global or if not declaring var makes it scoped in this mode. Based on not finding anything skimming through the examples, I assume the former?
I'm guessing the use of undeclared variables result in an error, instead of implicitly creating a global variable.
it also talks about the global object not being a place to add properties. So how you might do `window.foo = ...` or `globalThis.foo = ...` to make something from the local context into a global object. in this dialect I guess you would have to reserve any global objects you wanted to set with a `var` and then set them by reference eg
I think it means you can't assign to unbounded names, you must either declare with var for global, or let/const for local
This guy is incredible. Lzexe, Qemu, TinyCC to name just a few gems.
I wonder when does he have time to do those marvellous things
this would be a killer replacement for micro/circuitpython for embedded devices, assuming there's an elegant TS->MQJS transpile
It’s not even the languages or runtimes that inhibit embedded adoption but the software to hardware tooling. Loader scripts, HAL/LL/CMSIS, flashing, etc. They all suck.
It looks like if you write the acceptable MQJS subset of JS+types, then run your code through a checker+stripper that doesn't try to inject implementations of TS's enums and such it should just work?
I don't think you can transpile arbitrary TS in mqjs's JS subset. Maybe you can lint your code in such a way that certain forbidden constructs fail the lint step, but I don't think you can do anything to avoid runtime errors (i.e. writing to an array out of its bonds).
I find frustrating that imagination of very smart, talented and capable people is captured by JavaScript language, surely they could have done better.
Clarification added later: One of my key interests at the moment is finding ways to run untrusted code from users (or generated by LLMs) in a robust sandbox from a Python application. MicroQuickJS looked like a very strong contender on that front, so I fired up Claude Code to try that out and build some prototypes.
I had Claude Code for web figure out how to run this in a bunch of different ways this morning - I have working prototypes of calling it as a Python FFI library (via ctypes), as a Python compiled module and compiled to WebAssembly and called from Deno and Node.js and Pyodide and Wasmtime https://github.com/simonw/research/blob/main/mquickjs-sandbo...
PR and prompt I used here: https://github.com/simonw/research/pull/50 - using this pattern: https://simonwillison.net/2025/Nov/6/async-code-research/
Down to -4. Is this generic LLM-dislike, or a reaction to perceived over-self-promotion, or something else?
No matter how much you hate LLM stuff I think it's useful to know that there's a working proof of concept of this library compiled to WASM and working as a Python library.
I didn't plan to share this on HN but then MicroQuickJS showed up on the homepage so I figured people might find it useful.
(If I hadn't disclosed I'd used Claude for this I imagine I wouldn't have had any down-votes here.)
I think many subscribe to this philosophy: https://distantprovince.by/posts/its-rude-to-show-ai-output-...
Your github research/ links are an interesting case of this. On one hand, late AI adopters may appreciate your example prompts and outputs. But it feels like trivially reproducible noise to expert LLM users, especially if they are unaware of your reputation for substantive work.
The HN AI pushback then drowns out your true message in favor of squashing perceived AI fluff.
Yeah, I agree that it's rude to show AI output to people... in most cases (and 100% if you don't disclose it.)
My simonw/research GitHub repo is deliberately separate from everything else I do because it's entirely AI-generated. I wrote about that here: https://simonwillison.net/2025/Nov/6/async-code-research/#th...
This particular case is a very solid use-case for that approach though. There are a ton of important questions to answer: can it run in WebAssembly? What's the difference to regular JavaScript? Is it safe to use as a sandbox against attacks like the regex thing?
Those questions can be answered by having Claude Code crunch along, produce and execute a couple of dozen files of code and report back on the results.
I think the knee-jerk reaction pushing back against this is understandable. I'd encourage people not to miss out on the substance.
Counterpoint to the sibling comment: posting your own site is fine. Your contributions are substantial, and your site is a well-organized repository of your work. Not everything fits (or belongs) in a comment.
I'd chalk up the -4 to generic LLM hate, but I find examples of where LLMs do well to be useful, so I appreciated your post. It displays curiosity, and is especially defensible given your site has no ads, loads blazingly fast, and is filled with HN-relevant content, and doesn't even attempt to sell anything.
And again you're linking to your site. Maybe try pasting the few relevant sentences instead of constantly pushing your content in almost every comment. That's what people find annoying. Maybe link to other people's stuff more, or just write what you think here on HN.
If someone wants to read your blog, they will, they know it exists, and some people even submit your new articles here. There's no need to do what you're doing. Every day you're irritating more people with this behavior, and eventually the substance won't matter to them anymore, so you're acting against your own interests.
Unless you want people to develop the same kind of ad blindness mechanism, where they automatically skip anything that looks like self promotion. Some people will just see a comment by simonw and do the same.
A lot of people have told you this in many threads, but it seems you still don’t get it.
I'm determined to normalize linking to one's own writing, provided it's relevant to the conversation.
I think you're misreading what the "normalization" problem actually is and why my comment got a lot of upvotes.
You're not pushing against an arbitrary taboo where people dislike self links in principle. People already accept self links on HN when they're occasional and clearly relevant. What people are reacting to is the pattern when "my answer is a link to my site" becomes your default state, it stops reading like helpful reference and starts reading like your distribution strategy.
And that's why "I'm determined to normalize it" probably won't work because you can't normalize your way out of other people's experience of friction. If your behavior reliably adds a speed bump to reading threads forcing people to context switch/click out and wonder if they're being marketed to then the community will develop a shortcut I mentioned in my previous comment which basically is : this is self promo so just ignore.
If your goal is genuinely to share useful ideas, you're better off meeting people where they are: put the relevant 2-6 sentences directly in the comment, and then add something like "I wrote more about it on my blog" or whatever and if anyone is interested they will scroll through your blog (you have it in your profile so anyone can find it with one click) or ask for a link.
Otherwise you're not "normalizing" anything, you're training readers to stop paying attention to you. And I assure you once that happens, it's hard to undo, because people won't relitigate your intent every time. They'll just scroll. It's a process that's already started, but you can still reverse it.
No, I'm determined to normalize it. I would like a LOT more people to have personal websites where they write at length about things, and then share links to what they have already written where it is relevant to the conversation.
I'm actively pushing back against the "don't promote your site, don't link to it, restate your content in the comments instead" thing.
I am willing to take on risk to my personal reputation and credibility in support of my goal here.
There might be a bit of growth-hacking resistance here, and maybe some StackOverflow culture as well. Neither should be leveled at you, IMHO. I've followed and admired your work since Datasette launched, and I think you're exhibiting remarkably good judgment in how you discuss topics with links to deeper discussion, and it's in keeping with a long tradition of good practices for the web. Thanks for working to normalize the practice.
> can it run in WebAssembly?
You can safely assume so. Bellard is the creator of jslinux. The news here would be if it _didn't_.
> What's the difference to regular JavaScript?
It's in the project's README!
> Is it safe to use as a sandbox against attacks like the regex thing?
This is not a sandbox design. It's a resource-constrained design like cesanta/mjs.
---
If you vibe coded a microcontroller emulation demo, perhaps there would be less pushback.
Thank you for sharing.
A lot of HN people got cut by AI in one way or another, so they seem to have personal beefs with AI. I am talking about not only job shortages but also general humbling of the bloated egos.
> I am talking about not only job shortages but also general humbling of the bloated egos.
I'm gonna give you the benefit for the doubt here. Most of us do not dislike genAI because we were fired or "humbled". Most of us dislike it because a) the terrible environmental impacts, b) the terrible economic impacts, and c) the general non-production-readiness of results once you get past common, well-solved problems
Your stated understanding comes off a little bit like "they just don't like it because they're jealous".
It is because you keep over promoting AI almost every day of the week in the HN comments.
In this particular case AI has nothing to do with Fabrice Bellard.
We can have something different on HN like what Fabrice Bellard is up to.
You can continue AI posting as normal in the coming days.
Forget about the AI bit. Do you think it's interesting that MicroQuickJS can be used from Python via FFI or as a compiled module, and can also be compiled to WebAssembly and called from Node.js and Deno and from Pyodide running in a browser?
... and that it provides a useful sandbox in that you can robustly limit both the memory and time allowed, including limiting expensive regular expression evaluation?
I included the AI bit because it would have been dishonest not to disclose how I used AI to figure this all out.
It's interesting but I don't think it belongs as a comment under this post. I can use LLMs to create something tangential for each project posted on HN, and so can everyone else. If we all started doing this then the comment section will quickly become useless and not on point.
Offtopic but I went to your website and saw that you created hackernews-mute and recently I was commenting about how one must have created such an extension and ranted about it. So kudos for you to have created it earlier on.
Maybe we HN users have minds in sync :)
https://news.ycombinator.com/item?id=46359396#46359695
Have a nice day! Awesome stuff, would keep an eye on your blog, Does your blog/website use mataroa by any chance as there are some similarities even if they are both minimalist but overall nice!
I have something like that but as a userscript and with a toggle, it works pretty well for my needs.
Maybe someone finds it useful: https://paste.ubuntu.com/p/rD6Dz7hN2V/
Awesome stuff :D
Thanks for sharing it.
Thank you! I don't use Mataroa but I can see the similarities. My current blog setup is a Python script that parses content written in markdown and emits HTML. The CSS is inspired by the other minimal blogs I see here.
Thanks a lot for checking out my blog/project. Have a great day!
Software system is released, comments talk about how to integrate it with other software systems. Seems on-topic.
but there is signal in what people are inspired to do upon seeing a new project-- why not simply evaluate the interestingness level of these sorts of mashups on their own terms? it actually feels very "hacker"-y to go out and show people possibilities like this. i have no particular comment on how "interesting" the derivative projects are in this case, but i have a feeling if his original post had been framed more like "i think it's super interesting how easy it is to use via FFI on various runtimes X & Y (oh btw in the spirit of full transparency: i used ai to help me. and you can see more details at <link>). especially because i think everyone who peruses HN with some regularity is likely to know of simon's work in at least some capacity, and i am-- speaking only for myself-- essentially always interested in any sort of project he embarks on, especially his llm-assisted experiments and stuff. but hey-- at the end of the day, all of this value judgment is realized as plainly as possible with +1 / -1 (up- and down-vote) and i guess it just is what it is. if number bad, HN no like. shrug.
I agree that there is signal, and that phrasing his original post as you pointed out would have been better.
My issue is that the cost, in terms of time, for these experiments have really gone down with LLMs. Earlier, if someone played around with the posted project, we knew they spent a reasonable amount of time on it, and thus care about the subject. With LLMs, this is not the case.
Usually I watch your stuff very closely (and positively) because you're pushing the edges of how LLMs can be useful for code (and are a lot more honest/forthwright than most enthusiasts about it Going Horribly Wrong and how much work you need to do to keep on top of it.) This one... looks like a crossbar of random things that don't seem like things anyone would actually want to do? Mentioning the sandboxing bit in the first post would have helped a lot, or anything that said why that particular modes are interesting.
Yeah, I failed completely to explain the context here.
I'm currently on a multi-year side-quest to find safe ways to execute untrusted user-provided code in my Python and web applications.
As such, I pay very close attention to any new language or library that looks like it might be able to provide a robust sandbox.
MicroQuickJS instantly struck me as a strong candidate for that, and initial protoyping has backed that up.
None of that was clear from my original comment.
I had been in a similar boat and here are some softwares that I recommend or would discuss with you
https://github.com/libriscv/libriscv (I talked with the author of this project, amazing author fwsgonzo is amazing) and they told me that this has the least latency out of any sandbox at only minor consequence of performance that they know of
Btw for sandboxing, kvm itself feels good too and I had discussed it with them in their discord server when they had mentioned that they were working on a minimal kvm server which has since been open sourced (https://github.com/varnish/tinykvm)
Honestly Simon, Deno hosting/the way deno works is another good interesting tidbit for sandboxing. I wish something like deno's sandboxing capabilities came to python perhaps since python fans can appreciate it.
I will try to look more into your github repository too once I get more free.
Ah, reading this comment makes your original post 10x more interesting. I guess this is "start with why" in action. :)
It is depressing the age of llm coding power came during python and JavaScript.
Unfortunately it means those languages will be the permanent coding platforms.
> Unfortunately it means those languages will be the permanent coding platforms.
not really,
I suspect training volume has a role in debugging a certain class of errors, so there is an advantage to python/ts/sql in those circumstances: if, as an old boss once told me, you code by the bug method :)
The real problems I've had that hint at training data vs logic have been with poorly documented old versions of current languages.
To me, the most amazing capability is not the code they generate but the facility for natural language analysis.
my experience is that agent tools enable polyglot systems because we can now use the right tool for the job, not just the most familiar.
Simon although I find it interesting. And I respect you in this field. I still feel like the reason people call out AI usage or downvote in this case is that in my honest opinion, it would be also more interesting to see people actually write the code and more so (maintain) it as well and create a whole community/github project around microquickjs wasm itself
I read this post of yours https://simonwillison.net/2025/Dec/18/code-proven-to-work/ and although there is a point that can be made that what you are doing isn't a job and I myself create prototypes of code using AI, long term (in my opinion) what really matters are the maintainance and claim (like your article says in a way, that I can pin point a person whose responsible for code to work)
If I find any bug right now, I wouldn't blame it on you but AI and I have varying amount of trust on it
My opinion on the matter is that for prototyping AI can be considered good use but long term it definitely isn't and I am sure that you share a similar viewpoint.
I think that AI is so contrasting that there stops existing any nuance. Read my recent comment (although warning, its long) (https://news.ycombinator.com/item?id=46359684)
Perhaps you can build a blog post about the nuance of AI? I imagine that a lot of people might share a similar aspect of AI policy where its okay to tinker with it. I am part of the new generation and trust be told I don't think that there becomes much incentives long term unless someone realizes things of not using AI because using AI just feels so lucrative for especially the youngsters.
I am 17 years old and I am going to go into a decent college with (might I add immense competition to begin with) when I have passion about such topics but only to get dissuaded because the benchmark of solving assignments etc. are done by AI and the signal ratio of universities themselves are reducing but they haven't reduced to the point that they are irrelevant but rather that you need to have a university to try to get a job but companies have either freezed hiring which some point out with LLM's
If you ask me, Long term to me it feels like more people might associate themselves with hobbyist computing and even using AI (to be honest sort of like pewdiepie) without being in the industry.
I am not sure what the future holds for me (or for any of us as a matter of fact) but I guess the point I am trying to state is that there is nuance to the discussion from both sides
Have a nice day!
Your tireless experimenting (and especially documenting) is valuable and I love to see all of it. The avant garde nature of your recent work will draw the occasional flurry of disdain from more jaded types, but I doubt many HN regulars would think you had anything but good intentions! Guess I am basically just saying.. keep it up.
I didn't downvote you. You're one of "the AI guys" to me on HN. The content of your post is fine, too, but, even if it was sketch, I'd've given you the benefit of the doubt.
I think the people interacting with this post are just more likely to appreciate the raw craftsmanship and talent of an individual like Bellard, and coincidentally might be more critical of the machinery that in their perception devalues it. I count myself among them, but didn’t downvote, as I generally think your content is of high quality.
I downvoted because I'm tired of people regurgitating how they've done this or that with whatever LLM of the week on seemingly every technical post.
If you care that much, write a blog post and post that, we don't need low effort LLM show and tell all day everyday.
Here you go: https://simonwillison.net/2025/Dec/23/microquickjs/
No I mean post it as an HN post and if anybody cares to see it, they'll upvote that and comment in there. That, instead of pigging backing on other posts to get visibility.
I genuinely don't think it's interesting enough to warrant a top level post. It's interesting enough to be in a comment though!
I love it, I find the note interesting, educational, and adds to the discussion in context. Guess you're bound to get a few haters when you share your work in public, but I for one appreciate all your posts, comments, articles, open-source projects.
What is the purpose of compiling this to web assembly? What web assembly runtimes are there where there is not already an easily accessible (substantially faster) js execution environment? I know wasmtime exists and is not tied to a js execution engine like basically every other web assembly implementation, but the uses of wasmtime are not restricted from dependencies like v8 or jsc. Usually web assembly is used for providing sandboxing something a js execution environment is already designed to provide, and is only used when the code that requires sandboxing is native code not javascript. It sounds like a good way to waste a lot of performance for some additional sandboxing, but I can't imagine why you would ever design a system that way if you could choose a different (already available and higher performance) sandbox.
I want to build features - both client- and server-side - where users can provide JavaScript code that I then execute safely.
Just having a WebAssembly engine available isn't enough for this - something has to take that user-provided string of JavaScript and execute it within a safe sandbox.
Generally that means you need a JavaScript interpreter that has itself been compiled to WebAssembly. I've experimented with QuickJS itself for that in the past - demo here: https://tools.simonwillison.net/quickjs - but MicroQuickJS may be interesting as a smaller alternative.
If there's a better option than that I'd love to hear about it!
GraalVM supports running javascript in a sandbox with a bunch of convenient options for running untrusted code.
https://www.graalvm.org/latest/security-guide/sandboxing/
This is generally the purpose of JavaScript execution environments like v8 or jsc (or quickjs although I understand not trusting that as a sandbox to the same degree). They are specifically intended for executing untrusted scripts (eg web browsers). Web assembly’s sandboxing comes from js sandboxing, since it was originally a feature of the same programs for the same reasons. Wrapping one sandbox in another is what I’m surprised by.
Running v8 itself as a sandbox is non-trivial, at least embedded in a Python or Node.js or similar application.
The web is littered with libraries that half do that and then have a note in the README that says "do not rely on this as a secure sandbox".
Is it though? I have not personally used these libraries, but a cursory google search reveals several options: - cloudflare/STPyV8: [0] From cloudflare, intended for executing untrusted code. - Pythonmonkey: [1] Embeds spidermonkey. Not clearly security focused, but sandboxing untrusted code is literally the point of browser js engines.
It's a little less clear how you would do this from node, but the v8 embedding instructions should work https://v8.dev/docs/embed even if nodejs is already a copy of v8.
[0]: https://github.com/cloudflare/stpyv8 [1]: https://docs.pythonmonkey.io
... whoa, I don't know how I missed it but I hadn't seen STPyV8 before.
I'd seen PyV8 and ruled it out as very unmaintained.
One of my requirements for a sandbox is that it needs to me maintained by a team of professionals who are running a multi-million dollar business on it. Cloudflare certainly count! I wonder what they use STPyV8 for themselves?
... on closer inspection it doesn't seem to have the feature I care most about: the ability to constrain memory usage (see comment here https://github.com/cloudflare/stpyv8/blob/57e881c7fbe178c598...) - and there's no built-in support for time limits either, you have to cancel tasks from a separate thread: https://github.com/cloudflare/stpyv8/issues/112
PythonMonkey looks promising too: the documentation says "MVP as of September 2024" so clearly it's intended to be stable for production use at this point.
You should take a look at https://judge0.com/