[sunray]

I could also spend some money on a decent browser, sitting all day on Sunray which also have a dated browser sadly :(

"X multithread is re-enabled on Solaris 10.
If you have drag & drop X hang issue, update your Xserver or stay with 45.x.
If you are on Sun Ray, stay with 45.x."

[Raion]

Ideally any made for old PC browser will use motif or FLTK, as more advanced toolkits are a serious issue.

Because this code is Firefox backporting some fixes should be possible to improve the rendering capabilities. That still doesn't solve the JavaScript issue, though. Really we would need somebody who's an expert with MIPS ASM to build a JIT that could be crammed in.

[commodorejohn]

Classichasclass would be the one to talk to about that...guy probably knows more about doing JIT on vintage RISC workstations than anyone else.

[Raion]

Yeah, Cameron briefly expressed interest but he's really fixed on his Macs. Which is fine.

[Raion]

Update:

Reading through Firefox documentation, it appears the upper limit for possibly fixing a JIT for IRIX MIPSeb is Firefox 17, which is nearly 8 years old.

The reason for this is that FF18 defaults to a variety of Spidermonkey called "IonMonkey" which I've trawled through, and has no big-endian code in the versions covering 18-23 Firefox versions. So that effectively means any JIT efforts would need to be rendered from scratch.

Even if we have a MIPS JIT, something tells me SpiderMonkey's JS engine is probably too slow and heavyweight to be usable on all but the very last SGI machines. V8 and Webkit may prove similar, I'm afraid. I'm not sure of a good way around this.

[commodorejohn]

Again, I'd say hit up Cameron for advice even if he's not interested in picking up maintenance. PPC Macs are also big-endian, and TenFourFox is basically usable even on a mid-range G3.

Though, ultimately, I do think you are gonna run up against the fact that modern web design is so bloated that even an efficient browser will struggle to keep up with it on any hardware more than ten years old.

[sunray]

Update:

Reading through Firefox documentation, it appears the upper limit for possibly fixing a JIT for IRIX MIPSeb is Firefox 17, which is nearly 8 years old.

The reason for this is that FF18 defaults to a variety of Spidermonkey called "IonMonkey" which I've trawled through, and has no big-endian code in the versions covering 18-23 Firefox versions. So that effectively means any JIT efforts would need to be rendered from scratch.

Even if we have a MIPS JIT, something tells me SpiderMonkey's JS engine is probably too slow and heavyweight to be usable on all but the very last SGI machines. V8 and Webkit may prove similar, I'm afraid. I'm not sure of a good way around this.

But most RISC are bigendian eg Sparc and there is much later Firefox that 18 on Solaris? Why can not the Sparc version be use for a port?

Also maybe one way to sort out all dependencies in GTK and such would be to pick the oldest Solaris version with Java Desktop ie Gnome 2 and pick all it's GTK libs and tools from that version, then port to IRIX.

[Raion]

Because a JIT is architecture specific. The fact that SPARC has a JIT, as does PowerPC in TenFourFox **helps** but it doesn't solve the issue completely.

GNOME2 and GTK are kind of a lost cause for me, but of course, someone will probably allay those concerns eventually.

[pilot345]

Hey guys, first post here, but I am an old IRIX head from over on nekochan, happy to be part of wherever the tiny contingent of us end up on the vast internet..

Anyway, I have to share my opinion and ideas regarding this, because it is by far one of the biggest issues with IRIX usability long term.

Unfortunately with the way web development has become so bloated, and how browsers and accompanying software have exploded through versions and the developers care very little about anything but the latest hardware, I think that the actual task of porting a newer browser to IRIX is rather pointless, in terms of what we will actually be able to do with it, based on how fast it could run in even the best conditions and the fact that it will still be extremely old and not support a majority of websites.

I think the solution lies in offloading the browser work to a modern PC, and somehow displaying that rendering and making it usable on an IRIX desktop a la VNC or maybe something more specifically tailored, but accomplishing the same task. As far as I can surmise, this solution offers vastly greater promise, and long term usability than anything else. Of course we can do this now via VNC or similar, but if anyone has any ideas on how to improve this process and or maybe develop something more targeted that can do the task better, I would love to hear about it..

and with that I would love to hear any opinions all around this topic, as it is a heart wrenching one for us that still love to login to an IRIX desktop to work whenever we can..

[Raion]

Hi Pilot!

Glad to see you made your way here.

I think that the actual task of porting a newer browser to IRIX is rather pointless, in terms of what we will actually be able to do with it, based on how fast it could run in even the best conditions and the fact that it will still be extremely old and not support a majority of websites.

The goal at this point for me is HTML5/CSS3, and possibly limited JS support. But as I said, I find the options we can backport pretty unacceptable. Someone on the other forums has been working on ESR 17 of Firefox, and while it's possible to backport that, I don't think it'll be usable on anything but a Tezro or Onyx 350, which is self-defeating.

I think the solution lies in offloading the browser work to a modern PC, and somehow displaying that rendering and making it usable on an IRIX desktop a la VNC or maybe something more specifically tailored, but accomplishing the same task.

Remote X has hit a wall here. VNC is still too slow without JPEG SIMD support and "tight" encoding, which is a fancy way of saying good differential compression. Dodoid, one of the admins here, has been hitting a wall with that regarding his D1 project for the O2.

Personally, I only am willing to put money down (after the Inst bounty is fulfilled) on a browser that works acceptably on an R5k Indy or O2. Anything else is going to be unacceptable for me to put time/attention towards, or money. That being said, I may use it on my Tezro or Onyx2, I just think there's a colossal wasted potential with anything that isn't even remotely usable on a lower end system. The supply of high end SGIs seem to be drying up. 

One of the site side goals here has been to offer a gateway system for basic HTML capable browsers. I have a developer for it, I'm just holding off until we have the money and such to pay him.