[MPlayer-users] Re: (flame) FAQ entry concerning gcc 2.96

Alex Kanavin ak at cave.hop.stu.neva.ru
Mon Oct 22 18:39:17 CEST 2001

On Thu, 18 Oct 2001, Gabucino wrote:

>      Q:  What's the problem with gcc 2.96 ?                                     
>      A:  gcc 2.96 is RedHat's UNOFFICIAL
> True.

The name chosen for cvs snapshot was not a good one, agreed.

> 	 (it can be found only on RedHat sites,or RedHat distributions)
> True.

Mandrake (it is NOT Red Hat based).

> 	 and BUGGY gcc release.
> True.

Can you come up with an example of such a bug? Skipping mmx does NOT
count. Because gcc 3.x has exactly the same problem.

> 	 gcc 2.96 is TOTALLY unsupported by MPlayer,
> True.

There were TONS of reports from people happily using mplayer with gcc 2.96
and wondering why do you dislike it so much. Please point me to the
negative report - I've yet to see one.

> 	 because it simply SKIPS MMX/3DNow codes, it just does not compile it.
> True. 

This single statement is what needs to be corrected. The description of a
problem is highly inaccurate ("gcc 2.96 AND HIGHER skip ASSEMBLER codes
ONLY after SPECIAL condition which is a pipe in comments in inline
assembly" would be more accurate), and the problem itself is fixed.

>          Important: this is NOT an MPlayer-specific problem, numerous other
> 	 projects (DRI, avifile, etc..) have problems with this shit too.                                                    
> True.

Give me evidence, man! You mplayer developers always talk about problems 
with gcc 2.96, but never say what the problems actually are. I'd rather 
look into them than flame with you. And maybe even fix it, like I did with 
mmx code skipping. But I need a description of the ACTUAL PROBLEM, which 
you never come up with.

>          DO NOT USE gcc 2.96 !!!
> Good advice.

Well, based on incorrect assumptions.

>          If you really really want to do so, be sure to use the newest.         
> Good advice.

If so, why all the stupid childish barriers during compilation? This does 
nothing but damage your reputation. A simple RUN-TIME warning would be 

Anyway you guys seem to be way too stubborn on the issue. Can't you just 
accept that gcc 2.96 is not buggier than gcc 3.x? Or is this one of these 
"Red Hat is evil, we gotta blame them for something, even if it's entirely 
made up"? 

(now that I've let my steam off, I'm quickly putting on my asbestos


Homepage: http://www.sensi.org/~ak/

More information about the MPlayer-users mailing list