I love this. I deeply miss the productive, well-built UX era of OSX.
The screenshot shows a mix of styles. Some of it looks very old (eg the Finder analogue). Scrollbars are flat, traffic light window buttons are gel. Semi-puzzling so far in terms of UI design target.
We are on it. The whole UI is going to be overhauled a lot. Would you like to help us get things in shape? AppearanceMetrics.h in gershwin-eau-theme is a start.
I will have to try this out when I get the free time! I’ve been periodically checking on the GNUstep ecosystem since 2004, and this is the most exciting development I’ve seen since Étoilé from the late 2000s. Judging by the screenshots, the desktop appears to be Mac-like while also not being an exact clone.
If this desktop takes off, maybe we’ll finally see an ecosystem of applications that use GNUstep instead of GTK or Qt. In my opinion, the traditional charm of the Mac isn’t just the desktop; it’s the entire ecosystem of applications that conform to the Macintosh Human Interface Guidelines. It would be cool to finally see this happen with a GNUstep/Gershwin ecosystem!
Sadly, GNUStep is hardly moving, plus last time I remember there were still issues regarding the adoption of modern Objective-C, given that it is only supported on clang, and they were focused on using only what was available in GCC.
> “In my opinion, the traditional charm of the Mac isn’t just the desktop; it’s the entire ecosystem of applications that conform to the Macintosh Human Interface Guidelines”
Sadly this barely exists anymore.
Cross-platform Electron apps have replaced native AppKit. Cloud-based apps like Linear, Slack and Figma cater to the lowest common denominator of desktops by shipping their web client in a wrapper.
The last real native Mac app that was truly successful was probably Sketch ten years ago, and Figma ate their lunch.
Meanwhile Apple themselves have given up on the HIG. In the Alan Dye era, it’s been form over function across all the Apple operating systems. Their own apps don’t follow any guidelines and the latest macOS 26 is a UI disaster – probably the most inconsistent Mac release since OS X early betas.
I wholeheartedly and sadly agree with you. Seems like the idea of native apps on both macOS and Windows has been losing ground in favor of Electron apps. I understand the challenge of writing applications tailored to each platform and why Electron is so appealing to many companies, but I’d feel better about Electron if it were more conformant to platform HIGs and if it were less resource-intensive, especially now with RAM prices skyrocketing.
I hate how mainstream desktop computing has gone to crap in recent years. Thank goodness for free, open-source software.
I’m working on something similar for Linux. Would love to chat if this is interesting to you.
The idea is to bring the UX of OSX Snow Leopard back, adjusted for today’s possibilities (better developer experience, AI, etc.). I’m developing a DE, SwiftUI/AppKit-equivalent, and a bunch of reference apps I‘m personally missing in terms of quality (e.g. Raycast/Spotlight, Mail).
You would want to adjust it for today's display and input technologies. A high resolution OLED display deserves a different UI design than a 6-bit low-contrast TN LCD display did.
Interesting. I wonder what happened to Étoilé which was supposed to fill this niche and was chucking along fine until about 2015 or so. It got pretty far too with the underlying components.
Love this. Both my reason for wanting to replace macos and my biggest blocker to doing so is needing a desktop GUI with some HIG consistency, which is apparently a very old-fashioned idea these days.
Really, anything that can adhere to the old-school HIG well enough to offer A) consistent keyboard shortcuts across the apps that B) use a dedicated, thumb-actuated command key, I'm sold. (The `control` key, and thereby your pinkie finger, should both only be used for sending terminal escape sequences, as god intended [1].)
I am absolutely blown away to have the author of AppImage (Probono) partnering with me on a project I started that to be honest I didn't know if it would ever go anywhere. There is no better validation, and this is exactly what the spirit of creating open source solutions is all about for me. I wish I had known now years ago about GNUstep earlier. I took a path of learning FreeBSD deeper instead in 2004 when I started my career. I contributed to PCBSD as I learned programming in 2013 which eventually brought me to making FuryBSD for a short time.
When I started FuryBSD which was a livecd creator for FreeBSD that made it easy for others to spin up projects Probono noticed and started reaching out, helping me making some great contributions. It became the basis for the current GhostBSD LiveCD, HelloSystem, I believe RavynOS, and FyneDesk used it or at least were also using it in the past.
Probono blew me away with his work on LiveSTEP, and it just kind of stuck with me. I ended up silently carrying it forward, and getting back in touch after I realized what we could do with it. I recently gave Probono ownership for the GitHub org, and full creative control so I could focus on harder functional parts like a truly integrated WindowManager. It's all just been somewhat a miracle, and a matter of timing lining up I suppose. I am very much looking forward to seeing the cool things we can do together in 2026!
> I am absolutely blown away to have the author of AppImage (Probono) partnering with me
:-)
> which eventually brought me to making FuryBSD for a short time.
Ahaaaa. I did not realise that. Perhaps you should mention that in a FAQ or something? I think tying the different projects together like that would make it clear there is a quite considerable bit of history in here.
> I am very much looking forward to seeing the cool things we can do together in 2026!
Kudos for the positivity.
I left the GNUstep community a year or so back, after the admins got angry with me for daring to have opinions about the project that differ from theirs.
I think that as well as (1) a set of development libraries, it's also (2) a quite impressive set of apps, (3) an app packaging format, and perhaps most importantly (4) a quite complete desktop environment. They only seem to care about #1 and regard points 2-4 as annoying distractions.
For what it's worth, I know of two other active, current GNUstep-based desktop environments, which have slightly different focuses.
Apologies I am not familiar with how to quote back here. I'll just try to stay in order. I think that is a great suggestion to try to document the history of how Gershwin came to be. I'll put some thought into how to do that soon.
We do largely operate outside of GNUstep. Now we approach it like let us be the desktop, let them be the core libs. My take is I do think GNUstep should be marketed as more of a cross platform solution to build applications than anything else. You are more than welcome to come discuss ideas with us at Gershwin anytime.
GSDE (Screenshot.app), more so NextSpace a lot of things do not work with a lot of modifications on FreeBSD for example and I found the build systems unexpectedly difficult. I am a fan of the efforts otherwise and will try to make Gershwin components like WindowManager.app something they could use if they want to make use of in the future. I think each project has a place, and a role in promoting GNUstep. I wish they each had Live ISO's with installers. There is also agnostep now that looks promising by the way. https://github.com/pcardona34/agnostep
That was not intentional but just a coincidence actually. I came up with Gershwin as something to be comparable to "Darwin" as a core OS. I originally wanted to combine the Linux kernel with a Userland "familiar to switchers" more like a BSD and build on that. I also decided early on it was best to focus on being a DE that could run on anything and make the underlying OS not matter as much. Everyone involved really liked the name, so I went with it.
Screenshots look like OS X 1.0 and nothing like Rhapsody. I've found the OS X aesthetics unpleasant compared to how Rhapsody looked like so it was the final straw pushing me to Windows :)
The GTK theme engine from GNUstep can also be used to set a "Rhapsody" theme. It just allows using GTK themes. Here is an example of what that looks like https://github.com/pkgdemon/screenshots/blob/main/yellowbox-... I'd also like to make a native theme for that layout at some point.
I currently have the WindowManager.app I am fixing up that draws native decorations with GSTheme on to X11 windows. The screenshot in the gershwin desktop repo shows the result with chromium. I am also working on a Ladybird native GNUstep port where I need to fix the toolbar, rendering issues, and get the codebase in shape for a proper PR. Then I want to start working on fixing up an existing SwiftUI bridge implementation. This would also be a welcome contribution if someone can offer to contribute before I can eventually get to it. If that doesn't happen I would like to create a a native theme for this at some point.
Should be doable to put a Rhapsody theme on it... GNUstep is very flexible in this regard. Thanks to Method Swizzling, themes can change things pretty substantially.
Elementary OS does look superficially Apple-like, but on closer inspection, there's no menu bar, which is a core element of the macOS UI. So actually it's more like iPadOS on the desktop: apps generally only have hamburger menus, which personally I do not like at all, and it's limited by its insistence on Flatpak apps for everything. You can of course open a console and use `apt` to install whatever you want, but then the desktop quickly becomes less coherent and harmonious.
It's a good distro and I like it. It's easy, it has good accessibility, and as you say, it looks great. But I tried daily-driving it for a short time and found it too limiting for me. Once I'd manually installed Firefox, Thunderbird, Chrome, VLC, LibreOffice, Ferdium, Panwriter, VirtualBox and my other everyday tools, I wasn't using a lot of the distro's own tools any more, and suddenly the ones I was (settings, file manager, app launcher) became limiting.
I used elementary the past and appreciated the visual design. It just never went far enough for me by not having global menus, application bundles, etc. I think given a little more time here this year Gershwin will catch up so to speak. Then we will have not only how it looks, how it is designed, how it operates, how it runs on nearly anything, cross platform support, etc.
I recently began porting Ladybird from QT to GNUstep and the results are a 20 minute build on low powered raspberry pi vs 3 hours using most of the same APIs from AppKit. I believe I can take the same code, and make it build on Windows. I suppose this is my elevator pitch as to why.
I love this. I deeply miss the productive, well-built UX era of OSX.
The screenshot shows a mix of styles. Some of it looks very old (eg the Finder analogue). Scrollbars are flat, traffic light window buttons are gel. Semi-puzzling so far in terms of UI design target.
I found this: 10.2-era, pre-Brushed Metal, seems to be the target: https://github.com/orgs/gershwin-desktop/discussions/1
We are on it. The whole UI is going to be overhauled a lot. Would you like to help us get things in shape? AppearanceMetrics.h in gershwin-eau-theme is a start.
You know, I would, and I had already started looking into how to get a dev VM :)
Thankyou for the comment and polite interest, btw -- IMO an excellent way to encourage involvement in an open source project!
Looks like the Finder layout is inherited and may change: https://github.com/orgs/gershwin-desktop/discussions/8
This is pretty exciting!
I will have to try this out when I get the free time! I’ve been periodically checking on the GNUstep ecosystem since 2004, and this is the most exciting development I’ve seen since Étoilé from the late 2000s. Judging by the screenshots, the desktop appears to be Mac-like while also not being an exact clone.
If this desktop takes off, maybe we’ll finally see an ecosystem of applications that use GNUstep instead of GTK or Qt. In my opinion, the traditional charm of the Mac isn’t just the desktop; it’s the entire ecosystem of applications that conform to the Macintosh Human Interface Guidelines. It would be cool to finally see this happen with a GNUstep/Gershwin ecosystem!
Sadly, GNUStep is hardly moving, plus last time I remember there were still issues regarding the adoption of modern Objective-C, given that it is only supported on clang, and they were focused on using only what was available in GCC.
> “In my opinion, the traditional charm of the Mac isn’t just the desktop; it’s the entire ecosystem of applications that conform to the Macintosh Human Interface Guidelines”
Sadly this barely exists anymore.
Cross-platform Electron apps have replaced native AppKit. Cloud-based apps like Linear, Slack and Figma cater to the lowest common denominator of desktops by shipping their web client in a wrapper.
The last real native Mac app that was truly successful was probably Sketch ten years ago, and Figma ate their lunch.
Meanwhile Apple themselves have given up on the HIG. In the Alan Dye era, it’s been form over function across all the Apple operating systems. Their own apps don’t follow any guidelines and the latest macOS 26 is a UI disaster – probably the most inconsistent Mac release since OS X early betas.
This is why we think something inspired by the HIG needs to be reborn as open source.
I wholeheartedly and sadly agree with you. Seems like the idea of native apps on both macOS and Windows has been losing ground in favor of Electron apps. I understand the challenge of writing applications tailored to each platform and why Electron is so appealing to many companies, but I’d feel better about Electron if it were more conformant to platform HIGs and if it were less resource-intensive, especially now with RAM prices skyrocketing.
I hate how mainstream desktop computing has gone to crap in recent years. Thank goodness for free, open-source software.
I’m working on something similar for Linux. Would love to chat if this is interesting to you.
The idea is to bring the UX of OSX Snow Leopard back, adjusted for today’s possibilities (better developer experience, AI, etc.). I’m developing a DE, SwiftUI/AppKit-equivalent, and a bunch of reference apps I‘m personally missing in terms of quality (e.g. Raycast/Spotlight, Mail).
> adjusted for today’s possibilities
You would want to adjust it for today's display and input technologies. A high resolution OLED display deserves a different UI design than a 6-bit low-contrast TN LCD display did.
Definitely let's talk. You'll find us on GitHub Discussions and Libera Chat.
Is there a link for this project? It sounds exciting!
https://github.com/gershwin-desktop
Interesting. I wonder what happened to Étoilé which was supposed to fill this niche and was chucking along fine until about 2015 or so. It got pretty far too with the underlying components.
At least one of the lead developers of Étoilé moved on to CHERI.
For those wondering what CHERI is: https://cheriot.org/cheri/philosophy/isa/2025/11/19/cheri-or...
It's called Haiku now I think
Haiku is not related to Étoilé.
Haiku is an open source recreation of BeOS.
This was back in August, and at the time, I tried it and wrote about it:
https://www.theregister.com/2025/08/27/ghostbsd_2502/
Love this. Both my reason for wanting to replace macos and my biggest blocker to doing so is needing a desktop GUI with some HIG consistency, which is apparently a very old-fashioned idea these days.
Really, anything that can adhere to the old-school HIG well enough to offer A) consistent keyboard shortcuts across the apps that B) use a dedicated, thumb-actuated command key, I'm sold. (The `control` key, and thereby your pinkie finger, should both only be used for sending terminal escape sequences, as god intended [1].)
[1] https://en.wikipedia.org/wiki/Control_key#History
Gershwin is using the key next to the spacebar (whatever is written on it) as the Command key.
> Gershwin is a desktop environment based on GNUstep welcoming to switchers.
Might be just me, but that feels like an odd description. Assuming I'm understanding the intent correctly, perhaps something like
> Gershwin is a desktop environment based on GNUstep but made more intuitive for users coming from other desktops
might be easier to understand?
It's a reference to https://en.wikipedia.org/wiki/Switch_(advertising_campaign)
I LOVE this. There were a few other similar projects a few years back but they all seem to have stalled - hopefully this one will keep going
> There were a few other similar projects a few years back
Notably the Hello System, by Simon Peter, AKA probonopd.
I looked at that, too.
https://www.theregister.com/2023/01/31/hellosystem_08/
Probonopd is now involved with Gershwin and was quite active in its Github when I looked.
I am absolutely blown away to have the author of AppImage (Probono) partnering with me on a project I started that to be honest I didn't know if it would ever go anywhere. There is no better validation, and this is exactly what the spirit of creating open source solutions is all about for me. I wish I had known now years ago about GNUstep earlier. I took a path of learning FreeBSD deeper instead in 2004 when I started my career. I contributed to PCBSD as I learned programming in 2013 which eventually brought me to making FuryBSD for a short time.
When I started FuryBSD which was a livecd creator for FreeBSD that made it easy for others to spin up projects Probono noticed and started reaching out, helping me making some great contributions. It became the basis for the current GhostBSD LiveCD, HelloSystem, I believe RavynOS, and FyneDesk used it or at least were also using it in the past.
Probono blew me away with his work on LiveSTEP, and it just kind of stuck with me. I ended up silently carrying it forward, and getting back in touch after I realized what we could do with it. I recently gave Probono ownership for the GitHub org, and full creative control so I could focus on harder functional parts like a truly integrated WindowManager. It's all just been somewhat a miracle, and a matter of timing lining up I suppose. I am very much looking forward to seeing the cool things we can do together in 2026!
> I am absolutely blown away to have the author of AppImage (Probono) partnering with me
:-)
> which eventually brought me to making FuryBSD for a short time.
Ahaaaa. I did not realise that. Perhaps you should mention that in a FAQ or something? I think tying the different projects together like that would make it clear there is a quite considerable bit of history in here.
> I am very much looking forward to seeing the cool things we can do together in 2026!
Kudos for the positivity.
I left the GNUstep community a year or so back, after the admins got angry with me for daring to have opinions about the project that differ from theirs.
I think that as well as (1) a set of development libraries, it's also (2) a quite impressive set of apps, (3) an app packaging format, and perhaps most importantly (4) a quite complete desktop environment. They only seem to care about #1 and regard points 2-4 as annoying distractions.
For what it's worth, I know of two other active, current GNUstep-based desktop environments, which have slightly different focuses.
1. Ondrej Florian's GSDE: https://onflapp.github.io/gs-desktop/index.html
I've written a bit about this:
https://www.theregister.com/2023/07/06/two_new_debian_deskto...
I had great difficulty getting it to build on Debian 13, but I should try again at some point.
2. Sergii Stoian's NEXTSPACE: https://github.com/trunkmaster/nextspace
Slightly more mature but Stoian's been distracted recently by Russia invading his country, so it's not seen much work of late.
Apologies I am not familiar with how to quote back here. I'll just try to stay in order. I think that is a great suggestion to try to document the history of how Gershwin came to be. I'll put some thought into how to do that soon.
We do largely operate outside of GNUstep. Now we approach it like let us be the desktop, let them be the core libs. My take is I do think GNUstep should be marketed as more of a cross platform solution to build applications than anything else. You are more than welcome to come discuss ideas with us at Gershwin anytime.
GSDE (Screenshot.app), more so NextSpace a lot of things do not work with a lot of modifications on FreeBSD for example and I found the build systems unexpectedly difficult. I am a fan of the efforts otherwise and will try to make Gershwin components like WindowManager.app something they could use if they want to make use of in the future. I think each project has a place, and a role in promoting GNUstep. I wish they each had Live ISO's with installers. There is also agnostep now that looks promising by the way. https://github.com/pcardona34/agnostep
As vintagedave points out, there is a strange inconsistency of UI elements, but I look forward to seeing how this ends up.
Nice name, I suppose it recalls Apple’s Rhapsody project.
Additionally, Gershwin is the name of the intended follow-up to the cancelled Copland project.
That was not intentional but just a coincidence actually. I came up with Gershwin as something to be comparable to "Darwin" as a core OS. I originally wanted to combine the Linux kernel with a Userland "familiar to switchers" more like a BSD and build on that. I also decided early on it was best to focus on being a DE that could run on anything and make the underlying OS not matter as much. Everyone involved really liked the name, so I went with it.
Screenshots look like OS X 1.0 and nothing like Rhapsody. I've found the OS X aesthetics unpleasant compared to how Rhapsody looked like so it was the final straw pushing me to Windows :)
The GTK theme engine from GNUstep can also be used to set a "Rhapsody" theme. It just allows using GTK themes. Here is an example of what that looks like https://github.com/pkgdemon/screenshots/blob/main/yellowbox-... I'd also like to make a native theme for that layout at some point.
I currently have the WindowManager.app I am fixing up that draws native decorations with GSTheme on to X11 windows. The screenshot in the gershwin desktop repo shows the result with chromium. I am also working on a Ladybird native GNUstep port where I need to fix the toolbar, rendering issues, and get the codebase in shape for a proper PR. Then I want to start working on fixing up an existing SwiftUI bridge implementation. This would also be a welcome contribution if someone can offer to contribute before I can eventually get to it. If that doesn't happen I would like to create a a native theme for this at some point.
Should be doable to put a Rhapsody theme on it... GNUstep is very flexible in this regard. Thanks to Method Swizzling, themes can change things pretty substantially.
Why not use elementary OS which looks far better?
Elementary OS does look superficially Apple-like, but on closer inspection, there's no menu bar, which is a core element of the macOS UI. So actually it's more like iPadOS on the desktop: apps generally only have hamburger menus, which personally I do not like at all, and it's limited by its insistence on Flatpak apps for everything. You can of course open a console and use `apt` to install whatever you want, but then the desktop quickly becomes less coherent and harmonious.
It's a good distro and I like it. It's easy, it has good accessibility, and as you say, it looks great. But I tried daily-driving it for a short time and found it too limiting for me. Once I'd manually installed Firefox, Thunderbird, Chrome, VLC, LibreOffice, Ferdium, Panwriter, VirtualBox and my other everyday tools, I wasn't using a lot of the distro's own tools any more, and suddenly the ones I was (settings, file manager, app launcher) became limiting.
Exactly. Design, to us, is also how it works. Not just how it looks. The "feel" part in "look and feel".
Elementary/Pantheon resembles macOS only aesthetically. Functionally it’s similar to GNOME, which itself is closer to iPadOS than it is macOS.
Mac users looking to switch want the whole package: the aesthetics, the functionality, the design philosophy, and the holistic approach.
I used elementary the past and appreciated the visual design. It just never went far enough for me by not having global menus, application bundles, etc. I think given a little more time here this year Gershwin will catch up so to speak. Then we will have not only how it looks, how it is designed, how it operates, how it runs on nearly anything, cross platform support, etc.
I recently began porting Ladybird from QT to GNUstep and the results are a 20 minute build on low powered raspberry pi vs 3 hours using most of the same APIs from AppKit. I believe I can take the same code, and make it build on Windows. I suppose this is my elevator pitch as to why.