Hi Markus, The official FLAC source distribution comes with a test suite. Use "make check" - assuming you have something compatible with Unix compiler and build tools. If not, maybe you can look at the Unix make scripts and duplicate the same steps. From the README: The 'make check' step is optional; omit it to skip all the tests, which can take several hours and use around 70-80 megs of disk space. Even though it will stop with an explicit message on any failure, it does print out a lot of stuff so you might want to capture the output to a file if you're having a problem. Also, don't run 'make check' as root because it confuses some of the tests. Brian Willoughby Sound Consulting On Apr 13, 2008, at 10:36, Markus Ewald wrote: Because I'm currently running Vista x64 I thought I could try to do an x64 build of the Flac command line encoder. I used NAnt scripts instead of Visual Studio since Visual C++ Express 2008 doesn't support x64 builds (if anyone's interested, I can clean them up somewhat and send them to you). I took great care to use the exact same arguments for the compiler and linker as well as creating an x86 build in parallel to the x64 one. Now the funny part: to verify that my x64 build doesn't do anything fishy, I ran both my self-built x86 and my x64 encoder on a test song. The files ended up different. Just out of fun, I downloaded the official Win32 command line encoder and -- its output matched 100% with the x64 build! So it's actually my x86 encoder that's doing something different. Hm... Is there anything I can do to see why my 32 bits build generates different output from the official encoder binary? A test song or some kind of testing suite? -Markus-
Hi! Because I'm currently running Vista x64 I thought I could try to do an x64 build of the Flac command line encoder. I used NAnt scripts instead of Visual Studio since Visual C++ Express 2008 doesn't support x64 builds (if anyone's interested, I can clean them up somewhat and send them to you). I took great care to use the exact same arguments for the compiler and linker as well as creating an x86 build in parallel to the x64 one. Now the funny part: to verify that my x64 build doesn't do anything fishy, I ran both my self-built x86 and my x64 encoder on a test song. The files ended up different. Just out of fun, I downloaded the official Win32 command line encoder and -- its output matched 100% with the x64 build! So it's actually my x86 encoder that's doing something different. Hm... Is there anything I can do to see why my 32 bits build generates different output from the official encoder binary? A test song or some kind of testing suite? -Markus-
--- Markus Ewald <cygon at nuclex.org> wrote:> Now the funny part: to verify that my x64 build doesn't do anything > fishy, I ran both my self-built x86 and my x64 encoder on a test > song. > The files ended up different. Just out of fun, I downloaded the > official > Win32 command line encoder and -- its output matched 100% with the > x64 > build! So it's actually my x86 encoder that's doing something > different.that's not necessarily a problem. due to floating point differences between architectures, two builds can generate different but correct flac files. if they test with '-t' then they're fine. Josh ____________________________________________________________________________________ Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now. http://mobile.yahoo.com/;_ylt=Ahu06i62sR8HDtDypao8Wcj9tAcJ