Moving xlibs modules to xorg

Miles Bader miles.bader at necel.com
Sun Mar 5 22:46:47 PST 2006


Jamey Sharp <jamey at minilop.net> writes:
> Might another tool have been even better? I don't know: perhaps. Does
> some hypothetical improvement later justify continued pain now? I'll
> take the sure win now, thanks.

That's the thing:  everytime I've noticed a project try to do such a
move "properly" (widespread consultation, careful tabulation of all
the possible problems etc), it seems to result in a lot more pain
(huge flamewars) and no obvious advantage in the end.

I think basically every modern SCM these days (1) sucks in some
non-trivial way [exactly how often depends strongly on the user], and
(2) is about a zillion times better than CVS.

Having someone with half a clue just _pick_ something based on his
gut and hunker down while the resulting storm passes[*] seems as
good a method for getting out of the CVS hole as any.

[*] and it will almost certainly pass more quickly than the typical
painful and protracted process of selecting a new SCM in public.

> I still think that "monotone is slow" was a dumb reason for GIT to be
> created though.

In case you hadn't noticed, Linus is really, really, really a speed
freak.  I dunno if I like everything about git, but by gosh it is
fast...

[I use tla for my daily needs, and my #1 complaint about it is that
it has an absurdly lackadaisical attitude towards efficiency
... "Hey did I already stat that file?  Oh so what, I'll stat it a
few more thousand times, just in case!  Pipelining?  WHAT'S THAT?"]

-Miles
-- 
Do not taunt Happy Fun Ball.



More information about the xorg mailing list