st31276a wrote on 2026-05-05, 14:43:

The sse2 emulation might still be needed for things like firefox that is somehow difficult to compile without it.

For 586-class CPUs, modern versions of Firefox would run slow no matter what. You're better off with elinks/links2, dillo, or netsurf, all of which can be used without an X server and will be your best bet at doing any kind of fast/modern web browsing on a 586-based system.

mscdex wrote on 2026-05-05, 17:50:

st31276a wrote on 2026-05-05, 14:43:

The sse2 emulation might still be needed for things like firefox that is somehow difficult to compile without it.

For 586-class CPUs, modern versions of Firefox would run slow no matter what. You're better off with elinks/links2, dillo, or netsurf, all of which can be used without an X server and will be your best bet at doing any kind of fast/modern web browsing on a 586-based system.

Dillo got an update not long ago! 😃
I wonder if it's still possible to compile a new DOS version.
https://dillo-browser.github.io/release/3.2.0/

PS: PC GEOS got updated, too.
The latest version has a more recent web browser, even. 😎
Re: Retro OSes for retro computers

Jo22 wrote on 2026-05-05, 17:59:

Dillo got an update not long ago! 😃
I wonder if it's still possible to compile a new DOS version.
https://dillo-browser.github.io/release/3.2.0/

There was a 3.3.0 released on April 26.

mscdex wrote on 2026-05-07, 15:49:

Jo22 wrote on 2026-05-05, 17:59:

Dillo got an update not long ago! 😃
I wonder if it's still possible to compile a new DOS version.
https://dillo-browser.github.io/release/3.2.0/

There was a 3.3.0 released on April 26.

Hey, thank's a lot for the info! 😎👍
The more modern (maintained) web browsers there are out there, the better for the web.
Diversity is what keeps the web slowing down from getting too centralized too fast, I think.
The more rendering engines there are in circulation, the healthier the outcome.

the3dfxdude wrote on 2026-05-05, 13:17:

There are some guys working on a fork for the drivers, to bring some of them back.

Not really - all they're doing is fixing compiler warnings and ripping out XAA acceleration while not doing any conversion of the XAA code to EXA, or even testing the hardware. Good luck with using xf86-video-mach64 (covering everything from mach32 to Rage Pro/XL) and xf86-video-sis drivers as they will unconditionally segfault, my brief tests with xf86-video-mga and xf86-video-r128 were also not promising. A lot of the drivers are effectively framebuffers now, ironically one of the few drivers that has working EXA acceleration is xf86-video-trident, and it appears to be limited to Trident Blade3D 9880 cards.

So it's entirely possible to start fixing all the issues and bring back good support.

On the plus side - historical source code is easy to obtain and with the availability of datasheets and PCem/86Box (also MAME?) code, it shouldn't theoretically take too much effort for anyone who is willing and has the hardware. Going KMS is also a possibility, the Matrox G200 code in the kernel appears simple enough at a cursory glance that it shouldn't be too hard to port the code to other chipsets.

But if you wanted to do this for recent stuff, it's alot of work to base it on any mainstream distro, so you might as well start from scratch and make something work well on i586, or a known good variant, but which?

I already answered that in this thread - install Gentoo.