From: Takehiro T. <tak...@in...> - 2003-11-28 14:27:12
|
From: Gabriel Bouvigne <bou...@mp...> Subject: Re: [Lame-dev] [BUG] replaygain-accurate and FhG files Date: Fri, 28 Nov 2003 14:56:50 +0100 > Gabriel Bouvigne wrote: > > > It seems that our mpglib has some problems when used twice > > simultaneously. This needs to be investigated. > > Well, it seems that when using mp3 input and replaygain-accurate > simultaneoulsy, mpglib is using the same (global?) structures for both > decoding. Right. See mpglib_interface.c file in LAME4 branch... --------------------------------------------from here-------- #ifdef WITH_DMALLOC #include <dmalloc.h> #endif MPSTR mp; /* oops, not reentrant ... */ #ifndef NOANALYSIS plotting_data *mpg123_pinfo = NULL; #endif ------------------------------------------------------------- > It means that some work needs to be done to have proper isolation of > decoding data accross simultaneous calls. This would require some work > that I am not really willing to do now, so I will probably just disable > replaygain-accurate when input is mp3. Agree. > If someone would like to work on this problem, feel free to do it. I think we need completely new mp3 decoder for it, because mpg123 has many other bugs... Probabry replacing with libmad is good idea, but it will take too much time. IMHO It should be post LAME4. -- Takehiro TOMINAGA // may the source be with you! |