[Vorbis-dev] libvorbis 1.2.1 release?

Erik de Castro Lopo mle+la at mega-nerd.com
Mon Jun 2 17:03:38 PDT 2008


Note : I set my reply to headers to point to the list, so please
reply to the list and please to not CC me. Thank you.

Ivo Emanuel Gonçalves wrote:

> On 6/2/08, Erik de Castro Lopo <mle+la at mega-nerd.com> wrote:
> > Was libvorbis 1.2.1 ever released?
> 
> It's been put on hold until the situation in this ticket[1] is solved.
>  Maybe you can help.

<rant mode="goes to 11">

Unfortunately, I am intimately familiar with this "problem"
which is 100% due to Microsoft stupidity.

 1) Microsoft has *chosen* not to support the 1999 ISO C standard
    which has the lrint/lrintf/llrint/llrintf family of functions
    which solve this exact problem.

 2) Microsoft made the braindead decision not to support inline
    assembler when compiling for Win64.

As a short term solution, I support any of the patches discussed 
in the bug. However since I don't have a win64 machine, or use
windows compilers or windows itself for that matter, there is 
little I can do to move this issue forward.

For my own project libsndfile, I am now creating pre-compiled win32
binaries by cross compiling from Linux and running the libsndfile
test suite under Wine (the windows emulator (or not)). When cross
compiling I'm using GCC of course which supports the vast majority
of the C99 standard, including lrint etc. Furthermore, I do not
support any other way of compiling libsndfile for win32.

The Win64 version of the Linux to windows cross compiler is not
yet working well enough to compile libsndfile. However, when it
is working, that will be the only way to compile libsndfile for
win64 (where currently there is no way).

Plainly put, supporting compilers as out-of-date as the microsoft
ones is IMO, way to much of a PITA.

</rant>

Erik
-- 
-----------------------------------------------------------------
Erik de Castro Lopo
-----------------------------------------------------------------
Rule 1: Spammers lie.
Rule 2: If a spammer seems to be telling the truth, see Rule #1.


More information about the Vorbis-dev mailing list