search for: partword

Hi all, >From Monty's letter I know that ResVQTrain runs on vqd files that contain numbers separated by commas. I know that the .vqd files of this format generated by Vorbis compiled with RES_TRAIN contains partwords, i.e. codebook indeces assigned to partitions. I think that ResVQTrain should run on real data, not codebook indeces. Did I misunderstand something? Thanks Ervin <p><p>--- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsub...

...urrently under MPW/MrC. I can compile libogg/libvorbis fine (requires enabling the intrinsic alloca support), but there seems to be a problem with the code produced. As soon as real decode work starts, res2_inverse() dies because the decodemap it attempts to use has completely bogus values (so partword[l][k] is garbage, and the code faults on lookup into the secondstages array). I've been working on debugging this for a day or so, but I haven't found where it's actually screwing up. The optimization level makes not difference, which is strange. Presumably, this is a problem with t...