LWN.net Logo

Where will it *run*

Where will it *run*

Posted Jun 10, 2004 19:47 UTC (Thu) by Baylink (subscriber, #755)
Parent article: A look at Firefox 0.9

It's not exactly our "problem", per se, but there has always seemed to me to be a tendency in the F/OSS community not to impose arbitrary restrictions on such items as which processors a program will run on -- merely because we don't think it will be comfortable to use on slower machines.

If you want to run it on a 386, goes our notional argument, and you're willing to be that patient, go for it.

Comes now the milestone builds of Moz and FireFox.

Which will not run on my *586* laptop.

Someone decided that they should be built for P-II's and above; Pentium's need not apply.

I invite you to visualize my trying to build from source on my P-266 96MB laptop with KDE running, with only a couple hundred meg free on the 10 GB drive.

I invite you to imagine third-worlders trying to run it on their 486's.

I invite you to read the open bug , see all the comments and merged duplicates... and note that it's 4 months (and a couple of milestone builds) old, and that it hasn't even been assigned to the release manager yet.

Can This Project Be Saved?

Is anyone in the penthouse suite at mozilla.org listening?

Is it just me?

Cause, y'know, so many things are just me...


(Log in to post comments)

Where will it *run*

Posted Jun 12, 2004 22:20 UTC (Sat) by dvdeug (subscriber, #10998) [Link]

It's no longer true that most things will run on a 386. Proper atomic instructions need for threading only exist on 486 or above, so very few places make a 386 distribution any more.

Optimizing for a particular processor can often speed up the code. If I were running something much slower than my PIII-450, I'd find a faster browser, an older Netscape, or possibly Konqueror or Opera, or Lynx. Given that there are very few pre-Pentium II users left for Mozilla, taking the speed advantage over the compatibility with older chips seems to be reasonable.

Where will it *run*

Posted Jun 12, 2004 23:23 UTC (Sat) by Baylink (subscriber, #755) [Link]

> Given that there are very few pre-Pentium II users left for Mozilla,

My perception is that this line of thinking, which clearly is that of the Mozill build team as well, is
probably quite insular thinking; it may be true in the US (though I'm not even sure of that), but I
guarantee you there are easily 5 times as many Pentiums outside the US as there are Pentium-II+'s
inside the US....

Where will it *run*

Posted Jun 17, 2004 8:05 UTC (Thu) by dvdeug (subscriber, #10998) [Link]

I guarantee you there are easily 5 times as many Pentiums outside the US as there are Pentium-II+'s inside the US....

That's meaningless. Given that, there could still be fewer people percentage-wise using old computers outside the US then inside. In any case, I've never seen anybody come up with real numbers for either side of this debate. I do think your division is wrong, considering that several European nations and Japan have higher average incomes than the US, and many other nations are at least comparable. I still question that there's many Pentiums left. My father's Pentium 2 had a motherboard failure recently. How many Pentiums have failed and been replaced already? How many of the people running Pentiums are actively upgrading their computers and would try and run Mozilla?

I keep saying

Posted Jun 18, 2004 16:00 UTC (Fri) by Baylink (subscriber, #755) [Link]

"third-world" quite clearly, and everyone keeps not listening...

<sigh>

I keep saying

Posted Jun 18, 2004 22:37 UTC (Fri) by dvdeug (subscriber, #10998) [Link]

If you care to back and look at your last message, you didn't say third-world, unless you think "outside the US" means third-world. In any case, all you've done is make some unsubstatiated claims about the quantity and quality of machines in the third world. Nobody is going to be swayed by that alone.

You're right; it wasn't here I'd said it

Posted Jun 21, 2004 23:58 UTC (Mon) by Baylink (subscriber, #755) [Link]

On the other hand, your original reply didn't really speak to my assertions, either, and it isn't pertinent here: empirical testing has proven that building it to run on the 586 does not make it slower to run on the 686++ than building it so that it will *not* run on the 586 -- as the bugzilla comment chain shows.

Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds