: Re: [MPlayer-dev-eng] [PATCH] vf_spp: fix compiler-error and/or runtime crash

[Michael Niedermayer]

Hi

On Monday 03 January 2005 03:25, Martin Simmons wrote:
> >>>>> On Mon, 3 Jan 2005 00:34:24 +0100, "Dominik 'Rathann' Mierzejewski"
> >>>>> <dominik at rangers.eu.org> said:
> >
> > On Sunday, 02 January 2004 at 23:44, Michael Niedermayer wrote:
> > > Hi
> > >
> > > On Sunday 02 January 2005 23:18, Dominik 'Rathann' Mierzejewski wrote:
> > > > On Sunday, 02 January 2004 at 17:10, D Richard Felker III wrote:
> > > > > On Sun, Jan 02, 2005 at 11:57:47AM +0100, Michael Niedermayer wrote:
> > > > > > but OTOH, as the gcc developers dont consider things like this a
> > > > > > bug anymore (see my new sig ...) iam not sure how we should deal
> > > > > > with such things in the future
> > > > > >
> > > > > > [...]
> > > > > > --
> > > > > > Michael
> > > > > >
> > > > > > "In any case, just because code is syntactically "valid" GNU C
> > > > > > doesn't mean gcc can always compile it." -- justification to
> > > > > > close a gcc bug
> > > > >
> > > > > rotfl!!!
> > > > > i love it! :))
> > > >
> > > > Yeah, well, I don't believe everything I hear until I can verify it.
> > > > And I can't find this particular quote in any web archive, so please
> > > > either post a link or stop spreading FUD.
> > >
> > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11203
> >
> > Thank you. This is outrageous, even though it's only a comment from one
> > person. Especially when others show this is fixable. I wish this Steven
> > character could at least say "OK, we've screwed up, but we won't fix it
> > in 3.x". That'd be fair enough. But no, he has to go into the old "it's
> > your fault, stupid luser" routine. *sigh*
>
> He effectively does say that when he mentions GCC 4.0 or did you want
> grovelling?  

IMHO he does not: (quoting the IMHO relevant parts of the disscussion)
------
> It is crazy to do so.  Split up the inline-asm correctly.

fix gcc first so it doesnt load&store more then needed between the splited up 
parts
------
You've just constrained the compiler too much to do anything.  You're right 
that gcc should produce fewer loads and stores sometimes, but in this case 
I suggest you show that this actually hurts you still with GCC 4.0, I would 
hope it does better. In any case, just because code is syntactically "valid" 
------

so IMHO he says that gcc 4.0 will allow us to workaround the bug without 
reduced performance


> The quote about "syntactically valid" has been taken wildly 
> out of context, 

and which part of the comment would be not out of context? i cant quote the 
whole its too long


> because there are lots of syntactically valid but stupid 
> asm fragments that will break.  

stupid == everything gcc cant compile but finding a solution to the constrains 
is a matter of using any still available register for each operand


> They are like mencoder options :-). 

not really, if u submit a bugreport about some combination we will either 
explain why its not valid (violates spec, ...) or will keep the bug open or 
fix it

[...]
-- 
Michael

"In any case, just because code is syntactically "valid" GNU C doesn't 
mean gcc can always compile it." -- justification to close a gcc bug