Public SourceMage Discussion List

[Vlad Glagolev <stealth AT sourcemage.org>]

OK, let's discussion started!

> - this discussion should have taken place before the revert
Yes, this is the only place where I agree. My fault. I must start such thread 
a week before.

> - not everyone is affected, so the severity is not that CRITICAL
I'm not sure. Cause it depends on videocard, gtk+2 and xorg versions. So in 
life the
situations where everyone or nobody is affected are very rare. But as I said 
I don't wanna see
the users asking why theirs screen flickers (even 3, 2 or less). We must take 
care about EVERY
user, not the most. This is a policy taken not only in IT. I've researched 
the problem.

> - in your own words, most users use stable, so we should just make sure 
> this 
> is reverted in stable-rcs (until fixed), while test should stay updated

If the problem isn't solved by the original devs and it still occurs -- then 
test shouldn't stay
updated. Yes, there are users which stay with test. Yes, test is for testing. 
YES. BUT IT MUST
STAY USABLE (a.k.a. not broken).

> - if it is really such a huge problem, I'm sure the respective developers 
> will 
> fix it in short time
You sure? Hah! Let's see the stats. Problem occured in the beginning of July 
(as reported; or,
even earlier: 2.13.0 released on February 21).
That's about GTK+2 branch 2.13. And what do we see? Today's 10 Spetember 
2008. And what do we
see? Just a quote from http://bugzilla.gnome.org/show_bug.cgi?id=544072:
> Status:       UNCONFIRMED

Funny, eh? Seems GTK+2 devs don't use video-cards, lol. While xorg people 
invent/do SOMETHING,
gtk+ just don't care.. that's really funny. And this bug has pretty high 
level in xorg bugzilla
only because it was found by xorg developer (imho).

So, the key-quote from 
http://article.gmane.org/gmane.comp.freedesktop.xorg/30599:
> XRRGetScreenResources is an expensive call.  Always.  If gtk is
> calling it on every app startup they're absolutely insane.

And again. That's problem isn't so trivial as you think. To fix it, both API 
changes (I mean --
for both projects) are needed. See http://www.gnome.org/~ssp/randr/TODO for 
possible solutions.

Another key-quote from http://bugzilla.gnome.org/show_bug.cgi?id=544072:
> AIUI, this still needs:
> * implementation in xrandr drivers and server
> * new xrandr, server and driver releases

like this:
> So solutions I can think of here:
> a) request new API at xorg: anyone opened a bug for this?
> b) stop using the API altogether: what's the impact on gtk?
> c) use the API as little as possible, perhaps moving the information to a
> per-session location (e.g. xsettings); we could either move updating of this
> information to some other layer, or update it from whatever gtk app starts
> first, or periodically

So, I repeat. The problem isn't trivial, it's very complex.

And please, George. Read the whole threads, not only theirs titles.

Please understand. We shouldn't keep the broken version in -test, while GTK+ 
2.12 doesn't have
such problems and is pretty recent for now.

Thanks.

-- 
Dont wait to die to find paradise...
--
Cheerz,
Vlad "Enqlave" Glagolev

Attachment: pgpmaZuhO6y9l.pgp
Description: PGP signature