Sunday, July 27, 2008

Foxnews^H^H^H^Hconn conspiracy

What's up ass-captains? I've been very fortunate to be able to watch you guys shit your panties about this Foxconn thing, only to have the whole thing disputed. Leave it to the luser mob to create a whole movement out of some stupid bug.

But I'm not going to let you go that easily. You see, the saddest part is, you guys don't even realize how we all got into this whole mess in the beginning. OMGWTF! This motherboard's BIOS doesn't work w/Linux! Given that you know at least a Slashdot article's worth of information about how Linux distributions work, how did you expect it to work in the first place?

Seriously. Just think about it from Foxconn's perspective for two fucking seconds. What exactly are they supposed to do? If they want to test a motherboard with Linux, which Linux do they test it against? Redhat 9 or RHEL 3/4/5? Fedora 1-9? Ubuntu something? SUSE something? Debian something? Linus' tree? Con Kolivas' dead tree? Alan Cox's tree?

Even if they tested every 2.6.N release and released a board that supported them all, how are they to know that 2.6.N+1 won't break anything? Or that Redhat will apply some random patch that will break something. Or SUSE. Or Ubuntu. Why the hell do these guys deal with lusers at all?

Oh and it turns out the later versions of the kernel don't even claim to be Linux. Nice. (It's obvious why. Even if the OS does identify itself as Linux, which Linux would it be?) Hey guys, lets just forget about providing any kind of stable kernel behavior that hardware vendors can depend on. Instead, just claim we're Windows and emulate all the bugs! Luckily, then you can get into totally awesome issues like this one.

You may think that it's totally lame that HP has to work around a Vista bug like this, but at least they can. They can count on "Windows 2006 SP1" always meaning the same thing, and so they can fix problems for their users. The Linux kernel community has failed to provide stable targets for the rest of the industry to design around. They've made it extremely backwards and difficult for hardware vendors to fix things for Linux specifically, even if they wanted to.

I'm sure someone's gonna say, "well if they had designed it according to spec..." And I will respond, "obviously you've never tried to ship something that combines hardware and software before." If you create any non-trivial "spec" and hand a copy to hardware people and another to software people, then combine what they produce and ship it with no testing, I guarantee you that will be the end of your business.*

Of course, in the end, there's only one freetard answer to this whole mess. Open source the hardware design and BIOS implementation! Then even if there are bugs, the community will fix them eventually...

* And suffice it to say, the ACPI spec is thoroughly non-trivial.

blog comments powered by Disqus