[vorbis-dev] alloca() problem with CodeWarrior

[Firelight Technologies]

I guess I should clarify - I just mean 'if it means losing a kidney in
effort to work around, then don't do it, otherwise give it a go'.  
You cant bring up using old compilers, as that's obvious that using the
new compiler will probably fix it, but at the -present- point in time,
some compilers are deficient right -now-, and it might be quicker to do
something different, rather than wait for a future release of a compiler
(which might be a year away for example if it came from microsoft :) ).

 

> -----Original Message-----
> From: owner-vorbis-dev at xiph.org [mailto:owner-vorbis-dev at xiph.org] On
> Behalf Of Firelight Technologies
> Sent: Friday, 15 March 2002 12:58 PM
> To: vorbis-dev at xiph.org
> Subject: RE: [vorbis-dev] alloca() problem with CodeWarrior
> 
> I didn't mention compiler bug once.  I specifically mentioned support
of
> C library functions.  Compiler bugs are different and should be fixed
by
> the compiler manufacturer.
> 
> 
> Brett Paterson
> CEO / Lead Programmer.
> Firelight Technologies Pty Ltd.
> FMOD SoundSystem
> http://www.fmod.org
> 
> 
> > -----Original Message-----
> > From: owner-vorbis-dev at xiph.org [mailto:owner-vorbis-dev at xiph.org]
On
> > Behalf Of Segher Boessenkool
> > Sent: Friday, 15 March 2002 12:50 PM
> > To: vorbis-dev at xiph.org
> > Subject: Re: [vorbis-dev] alloca() problem with CodeWarrior
> >
> > > I agree with brian, if something didn't work on one pc out of a
> > > thousand, I would try and work around it,
> >
> > Ah, like this:
> >
> > -- the user might want to use gcc-2.95 on x86 with an OS distro that
> >    thinks all the world is an intel pentium pro/2/iii
> >    ==>  no fp comparisons, again.
> >
> > -- the user might want to use gcc-2.91.66 on x86
> >    ==>  we can't reliably use fp comparisons.
> >
> > -- the user might want to use gcc-3.0.2 on ppc
> >    ==>  we can't use floating point comparisons to 0.f
> >
> > -- the user might want to use MSVC older than, erm, 5 I think
> >    ==>  we can't reliably use floats in static initializers.
> >
> > -- the user might want to use some older CW version
> >    ==>  we must make sure all our structs and stack parameter frames
> >         are 16-byte aligned, in case the user has enabled some goofy
> >         compiler options and expects the resulting lib to play nice
> >         with other libs.
> >
> > Just some of the compiler bugs I noticed while hacking Vorbis (well,
> #2
> > wasn't discovered by me, and #5 was a well-known bug already.  And
> there
> > already was a compiler fix for it).
> >
> > All of these happen more than 1 in 1000 users, and none of these can
> > be worked around without huge sacrifices in either code speed or
> > legibility or maintainability or whatever.
> >
> >
> > Yes, we don't use C99 features, because not every compiler has it.
> > But we won't go as far as writing everything K&R compatible.
> > Whether or not to use alloca() is somewhere inbetween.
> >
> >
> > Segher
> >
> >
> > --- >8 ----
> > List archives:  http://www.xiph.org/archives/
> > Ogg project homepage: http://www.xiph.org/ogg/
> > To unsubscribe from this list, send a message to 'vorbis-dev-
> > request at xiph.org'
> > containing only the word 'unsubscribe' in the body.  No subject is
> needed.
> > Unsubscribe messages sent to the list will be ignored/filtered.
> 
> 
> --- >8 ----
> List archives:  http://www.xiph.org/archives/
> Ogg project homepage: http://www.xiph.org/ogg/
> To unsubscribe from this list, send a message to 'vorbis-dev-
> request at xiph.org'
> containing only the word 'unsubscribe' in the body.  No subject is
needed.
> Unsubscribe messages sent to the list will be ignored/filtered.

<p>--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.