- Welcome to NeXTcommunity.
Recent posts
#61
Hardware / Re: Flashing LED -NeXTStation ...
Last post by marvin - Sep 30, 2026, 06:36 PMEven though you have a color slab, maybe Brian Archer's recap guide http://www.asterontech.com/Asterontech/nextstation_mono_retore.htm is still helpful.
#62
Software / Re: Releasing the Radeon 9250 ...
Last post by onionmixer - Sep 30, 2026, 06:15 AMQuote from: Rhetorica on Sep 29, 2026, 03:29 PMAbsolutely glorious! Making Quake usable on NeXT is very poetic, since development and even mapping happened on our beloved operating system. (We have the source for QuakeEd here. They hadn't even moved on to the OPENSTEP API yet.) Rhapsody did eventually get binaries for Quake 2 and Q3Test, but they were ported by OmniGroup. (I'm not clear on whether the Q2 binary supported Intel or not, it may have been PowerPC only.)
All of this is to say... if you send a screenshot to John Carmack or John Romero, they might be pleasantly surprised.
https://github.com/onionmixer/openstep-matrox-remade
I've already posted this on the .org community, but... I've actually done a similar project before with the Matrox G450 PCI. I implemented OpenGL acceleration using Matrox's Warp. This new project is sort of an extension of that.
Back in the Rhapsody days, OS X had already started supporting OpenGL at the OS level. I also remember the news back then about the OmniGroup quickly porting Quake. However, OPENSTEP didn't provide a way to support OpenGL at the OS level, so when I rewrote the Matrox driver, I implemented it by mapping the VGA's VRAM directly to AppKit. The basic theory for this new Radeon driver is the same. (Though it's a secret that when I thought of this approach and presented it to AI, I actually got praised by the AI, haha)
OPENSTEP doesn't provide a display overlay at the OS level, and if I had tried to implement OpenGL purely through hardware overlays, I would have had to deal with a lot more headaches regarding how the OS handles keyboard and mouse input. Plus, windowed mode probably wouldn't have been feasible. I wanted to focus as much as possible on portability, which is why I ported SDL2 to OPENSTEP and used `sdquake` as a proof-of-concept for OpenGL on OPENSTEP.
I know that QuakeEd originally started on OPENSTEP, and making Quake *itself* run on OPENSTEP—rather than just Ed—was one of my long-held bucket list goals. Since I already pulled it off with the Matrox, the thrill is maybe just a tiny bit less this time around (smile), but thanks to the hardware performance, the Radeon 9250 PCI gets much better framerates
As a side note... a friend of mine posted a video of the Quake running on the Matrox setup on X and retweeted John Carmack, but judging by the lack of any reaction... I'm guessing Quake is just a distant memory to John Carmack by now
#63
Software / Re: New apps - SSH client, IRC...
Last post by ptek - Sep 30, 2026, 01:36 AMQuote from: jade_nekotenshi on Sep 26, 2026, 03:45 AMI'm not sure how the community feels about AI-assisted coding
AI-assisted coded software for NS/OS is better than no software for NS/OS
Looking forward to trying the SSH software on OS 4.2 patched with VirtualBox.
#64
Software / Re: New apps - SSH client, IRC...
Last post by jade_nekotenshi - Sep 29, 2026, 11:47 PMStepSSH 1.4.1 is now available. Tested on all three architectures now.
Big changes:
Tested and working on SPARC
Cipher selection box
Big optimization pass that has increased bulk throughput considerably.
Big changes:
Tested and working on SPARC
Cipher selection box
Big optimization pass that has increased bulk throughput considerably.
#65
Software / Re: New apps - SSH client, IRC...
Last post by itomato - Sep 29, 2026, 04:26 PMQuote from: jade_nekotenshi on Sep 26, 2026, 03:45 AMI'm not sure how the community feels about AI-assisted coding, but...
Personally, I am for it.
#66
Marketplace / Re: NeXT eBay Finds!
Last post by jade_nekotenshi - Sep 29, 2026, 04:21 PMQuote from: ZombiePhysicist on Sep 29, 2026, 02:08 PMhttps://ebay.io/m/rjsDD6
HP Apollo runs nextstep 3
Yep, the 715 is specifically called out as compatible on the box!
#67
Software / Re: New apps - SSH client, IRC...
Last post by jade_nekotenshi - Sep 29, 2026, 04:18 PMThis is what Claude had to say:
I'm not sure if this is a bug in the 68040's FPU or a bug in the compiler, or maybe even a bug in Previous, but it's kluged around now!
QuoteIt's a compiler bug, and I've fixed it in both StepSSH and RatChat. It's untested on the 68040 so far. gcc 2.7.2's m68k backend miscompiles a 64-bit right shift by the constant 16, and the result comes back with a zero high word. car25519 does exactly that shift, so every field multiply was wrong. That broke X25519 and Ed25519 and nothing else.
I don't have to guess that this explains it. I made the host reproduce the bug by zeroing the high word of that shift in a scratch copy of nacl.c. All 29 wrong lines from your 68040 came out identical, including the X25519 outputs, the Ed25519 key and signature, and the raw carry limbs. Nothing was left unexplained.
I'm not sure if this is a bug in the 68040's FPU or a bug in the compiler, or maybe even a bug in Previous, but it's kluged around now!
#68
Software / Re: Releasing the Radeon 9250 ...
Last post by Rhetorica - Sep 29, 2026, 03:29 PMAbsolutely glorious! Making Quake usable on NeXT is very poetic, since development and even mapping happened on our beloved operating system. (We have the source for QuakeEd here. They hadn't even moved on to the OPENSTEP API yet.) Rhapsody did eventually get binaries for Quake 2 and Q3Test, but they were ported by OmniGroup. (I'm not clear on whether the Q2 binary supported Intel or not, it may have been PowerPC only.)
All of this is to say... if you send a screenshot to John Carmack or John Romero, they might be pleasantly surprised.
All of this is to say... if you send a screenshot to John Carmack or John Romero, they might be pleasantly surprised.
#69
Marketplace / Re: NeXT eBay Finds!
Last post by ZombiePhysicist - Sep 29, 2026, 02:08 PM #70
General Discussion / Re: Thoughts on AI usage for r...
Last post by jeffburg - Sep 29, 2026, 02:00 PMThanks for bringing this up. As you know from my post on my AI Assistant I made for iOS 5+ and PowerPC Macs, I am deeply into AI development. I am a professional software developer, so my starting point is a bit different, but I would say at this point I would classify myself as AI pilled. So take what I say with a grain of salt. But I feel like DHH getting up in front of 1000 Ruby Developers and ranting about how I am rewriting everything in rust because its faster and better and I don't need to look at the terribly ugly code (reference). I want to be clear, I understand both AI and DHH are controversial and I am not a superfan of all aspects of either of them. With AI particularly, the licensing of code seems dubious. As well energy usage is a problem. But I just can't ignore what it gets for me.
I am now writing toy apps for my favorite retro devices in no time at all. And because of AI I am writing them in pure C (except the UI which is obviously my original love affair, Objective-C). But the important part is that I am writing the highest performance, most reliable, most "sophisticated" code I have ever written in my life. By sophisticated I mean that I add in nice quality of life features all the time I would never write myself on a side project because it's just too much work. The revelation is so great that at this point, I am able to make my iMac G4 (2003) and my iPhone 5 (2012) actually usable in 2026 with all the things I need to do. They all have:
I write about all of this in my unenshittification series on my blog. But basically, I am unenshittifying my digital life and having a ton of fun doing it. That said, I have not been pushing it very hard online because the retro tech community seems to be pretty averse to AI... and honestly I just don't get much engagement. That does not bother me as I am building things for myself but I really think the retro tech community should try to embrace it (despite obvious concerns)
I think the reason is that for retro software... no one is going to make any money, so there is no incentive to write things. It's pure passion. And in that case having this acceleration and being able to step out of your comfort zone are so critical. Like for me, I am not a C developer and I don't want to become one. But with AI, I can write thousands of lines of C code in a few days... when I look at it, it looks fine. It's so simple, C is such a simple language that I can easily see what it is doing and that it's doing it in a simple way. And I can see its doing memory management correctly (at glance). I am not an expert so I can't tell if it's doing a buffer overflow or something. That part is just determined by the static analyzer and using the app.
But thats the other part about retro software development. The license doesn't matter, it's all free and open source, offered without warranty. So if something breaks or loses data, well, sorry I tried my best but its a free app I wrote in my free time and you should not use a 20+ year old computer in production anyway.
Lastly, is the speed of acceleration. When I started this journey in January, it was pretty slow and painful to write the code with AI. But now, 6-8 months later... honestly, I actually never tell the AI what to do. I tell it my idea, ask it what it thinks and ask it how we can do the most simple and high performance implementation of that thing... and after piling these up after many small small sessions, you have a high performance app that does modern things on ancient computer. And thats because the simple fact is that 99% of modern apps just download JSON files and then display them in a list.... thats Twitter, Instagram, Ebay, WhatsApp, etc. You don't need a supercomputer for this task. It's easy for the CPU. The reason these apps are so slow on our insanely fast modern supercomputers is because they are just full of slop code... not AI slop... purely human created Slop code that existed long before AI existed.
So yeah, again, I apologize for my rant. But I just absolutely love AI for retro software development and have no plans to stop... even if sharing it in our communities is totally taboo. I hope it's not and I hope we can all learn to embrace it together. But I think it's one of the greatest things that could happen to retro computing.
I think we are only 1 year away or so from a total novice from being able to write a modern app on a modern device from a simple prompt. Retro apps are harder to write, so maybe that same milestone is 2 years away on retro apps. But I think we are not far from enabling anyone to make the apps they want for their favorite and most beloved retro devices and I am super excited about it.
I am now writing toy apps for my favorite retro devices in no time at all. And because of AI I am writing them in pure C (except the UI which is obviously my original love affair, Objective-C). But the important part is that I am writing the highest performance, most reliable, most "sophisticated" code I have ever written in my life. By sophisticated I mean that I add in nice quality of life features all the time I would never write myself on a side project because it's just too much work. The revelation is so great that at this point, I am able to make my iMac G4 (2003) and my iPhone 5 (2012) actually usable in 2026 with all the things I need to do. They all have:
- iCloud Email, Contacts, Calendar Sync
- Youtube video download and playback
- AI Personal Assistant
- Social Media (Mastodon)
- iMessage Messaging
- LINE Messaging
- News (RSS+The Old Reader)
- Web Article Viewing (thanks to the reader proxy from next community actually)
- And more is coming
I write about all of this in my unenshittification series on my blog. But basically, I am unenshittifying my digital life and having a ton of fun doing it. That said, I have not been pushing it very hard online because the retro tech community seems to be pretty averse to AI... and honestly I just don't get much engagement. That does not bother me as I am building things for myself but I really think the retro tech community should try to embrace it (despite obvious concerns)
I think the reason is that for retro software... no one is going to make any money, so there is no incentive to write things. It's pure passion. And in that case having this acceleration and being able to step out of your comfort zone are so critical. Like for me, I am not a C developer and I don't want to become one. But with AI, I can write thousands of lines of C code in a few days... when I look at it, it looks fine. It's so simple, C is such a simple language that I can easily see what it is doing and that it's doing it in a simple way. And I can see its doing memory management correctly (at glance). I am not an expert so I can't tell if it's doing a buffer overflow or something. That part is just determined by the static analyzer and using the app.
But thats the other part about retro software development. The license doesn't matter, it's all free and open source, offered without warranty. So if something breaks or loses data, well, sorry I tried my best but its a free app I wrote in my free time and you should not use a 20+ year old computer in production anyway.
Lastly, is the speed of acceleration. When I started this journey in January, it was pretty slow and painful to write the code with AI. But now, 6-8 months later... honestly, I actually never tell the AI what to do. I tell it my idea, ask it what it thinks and ask it how we can do the most simple and high performance implementation of that thing... and after piling these up after many small small sessions, you have a high performance app that does modern things on ancient computer. And thats because the simple fact is that 99% of modern apps just download JSON files and then display them in a list.... thats Twitter, Instagram, Ebay, WhatsApp, etc. You don't need a supercomputer for this task. It's easy for the CPU. The reason these apps are so slow on our insanely fast modern supercomputers is because they are just full of slop code... not AI slop... purely human created Slop code that existed long before AI existed.
So yeah, again, I apologize for my rant. But I just absolutely love AI for retro software development and have no plans to stop... even if sharing it in our communities is totally taboo. I hope it's not and I hope we can all learn to embrace it together. But I think it's one of the greatest things that could happen to retro computing.
I think we are only 1 year away or so from a total novice from being able to write a modern app on a modern device from a simple prompt. Retro apps are harder to write, so maybe that same milestone is 2 years away on retro apps. But I think we are not far from enabling anyone to make the apps they want for their favorite and most beloved retro devices and I am super excited about it.