[linux-elitists] on hardware compatability

Nick Moffitt nick@zork.net
Thu Sep 19 15:23:42 PDT 2002


So I was reading someone's installation woes over getting the W2k
problem installed on his thinkpad (Red Hat installed instantly, he
said, and allowed him to then download a complicated constellation of
patches to w2k), and remembering a story that Maddog used to tell.

Back when, you'll remember that he worked for the linux publicity
nonprofits, just playing evangelist.  He worked at
DEC^WDigital^WCompaq^HP, and used to tell stories about how the Ultrix
engineers claimed that shared libs were too hard to impliment but
someone submitted a shlib patch to Linux after a weekend of work.  He
was one of those first folks to say "we can sneak this stuff into big
companies, and sell it to them when they find it".

ANyway, he said that one trick he used to do with hardware vendors
(initially Compaq) was to ask for time with the hardware muckety-mucks
and one of their top-of-the-line bread-and-butter server boxes.  He'd
say that he was demoing Linux, just to show them what the hobbyist
community was doing with PCs, like expand your horizons and find new
markets and blah blah blah.

He'd show up at the meeting with a grubby-as-sin white-box PC salvaged
from some back lot or something.  Like, dusty and grimy and ugly.
He'd plug in this two-year-old Pentium or whatever, and walk them
through the install, showing them how straightforward it was.

Then he'd pop up their own box, and the installer would fail
invariably on some drive controller or BIOS looneyness or something
like that.  He'd say "See?  Even this stupid old piece of dirty crap
can do this, while your shiny box with all the chrome just breaks. And
why?  Because you're ignoring this OS!"

Anyway, so it started me thiking about the contrast between the Maddog
demo and this guy's thinkpad woes.  A lot of that crap has really been
solved now.  I mean, yeah there are still goofy incpatabilities here
and there, but the Linux kernel doesn't lag behind the hardware so
badly any more.  Some of that is because of more developers, or bette
rvendor relations, or whatever, but I think I have another explanation
for it.

The bust killed a lot of hardware development.

I mean, the kernel folks were struggling to keep up with the
insatiable pace of dot-com-boom hardware development.  These compaq
servers and new RAID controllers were so strange because they were
being built all licketysplit to be installed in the next wave of
chump-buy Web servers.  

But that's all slowed down.  We no longer have the video-card-per-day
phenomenon.  Lots of vendors have died out and the driver authors
don't need to move as fast to keep up.  So we see instead a next-rev
of the neomagic chipset, with minimal work for both the neomagic folks
and the XFree86 authors.  

Maybe I'm missing out.  Maybe there are scores of new boxes out there
with miserable compatability.  Maybe some gamer will pipe in and say
"No way man, there are like four new companies with unsupported 3-D
cards that do things other cards never tried to!".  It just seems to
me like the kernel is catching up because hardware is taking a hare's
nap.

-- 
A: No.
Q: Should I include quotations after my reply?



More information about the linux-elitists mailing list