--- Pisco <pisco@mac.com> wrote:> This may clarify a few things. I compiled flac from the source > package > on flac.sourceforge.net since the Darwin package did not contain the > XMMS > plugin. I compiled it using the ./configure, make, make all > sequence, > and it compiled fine. > When the flac plugin is alone in the XMMS Input directory, it works > correctly and shows up correctly in the Input panel of the XMMS > preferences. (i.e. "FLAC Player v1.0.2 [libxmms-flac.so]") When the > shn > plugin is alone in the XMMS input directory, it works correctly and > shows > up correctly in the Input panel. (i.e. "SHN Player 2.2.5 > [libshn.so]") > When they are both present in the input directory, things start to > get > screwy. If they are both named appropriately, the shn plugin works, > the > flac plugin doesn't, and the shn plugin shows up twice in the Input > panel. (i.e. "SHN Player 2.2.5 [libxmms-flac.so]\nSHN Player 2.2.5 > [libxmms-flac.so]") It seems to load the first of the two it sees, > but > take the filename from the second. If I rename the shn plugin to > load > after the flac plugin (i.e. to libxmms-shn), then the opposite occurs > - > the flac plugin loads correctly, and the listings show two flac > plugins. > (i.e. "FLAC Player v1.0.2 [libxmms-shn.so]\nFLAC Player v1.0.2 > [libxmms-shn.so]")Ah, I see. My first guess at what is happening is that both plugins are stomping on each other's global symbols. Looking at flac/src/plugin_xmms/plugin.c, it looks like the only one it could be is get_iplugin_info(): InputPlugin *get_iplugin_info() { flac_ip.description g_strdup_printf("FLAC Player v%s", FLAC__VERSION_STRING); return &flac_ip; } I copied the basic plugin design from another plugin (I think it was XMMS' WAVE reader). I'm not familiar enough with XMMS' plugin implementation yet but looking at it I don't see how the get_iplugin_info() design can work with multiple shared libraries. I CC'ed the flac-dev list to see if anyone knows off the top of their head. Josh __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com
On Tue, Jun 11, 2002 at 10:16:11AM -0700, Josh Coalson wrote:> Ah, I see. My first guess at what is happening is that both plugins are > stomping on each other's global symbols. Looking at > flac/src/plugin_xmms/plugin.c, it looks like the only one it could be is > get_iplugin_info(): > [...] > I copied the basic plugin design from another plugin (I think it was XMMS' > WAVE reader). I'm not familiar enough with XMMS' plugin implementation > yet but looking at it I don't see how the get_iplugin_info() design can > work with multiple shared libraries. I CC'ed the flac-dev list to see if > anyone knows off the top of their head.It dlopen()s the shared object and uses dlsym() to find the get_iplugin_info which is in that particular object, so there should not be a problem with conflicts. This could be a problem with whatever xmms build is being used, or (somewhat less likely) a dynamic linker bug. -- - mdz
Hello, This is an inquiry which could be answered by looking in the code (which I will do), but a direct answer by the expert would be interesting and useful. Given a similar musical content (i.e. same dynamic range and frequency response) encoded in both 24/96 and 16/44.1 formats, will flac produce roughly the same size compressed file for both formats? I'm wondering whether it's valid to draw an analogy between image files and sound files. You can scan a sheet of text and store it both medium-resolution 8-bit uncompressed TIF and 32-bit high-resolution uncompressed TIF, but since the content is the same, they'll both compress down to (roughly) the same size. Paul
--- Paul Kubicz <pkube@umich.edu> wrote:> Hello, > > This is an inquiry which could be answered by looking in the code > (which > I will do), but a direct answer by the expert would be interesting > and > useful. > > Given a similar musical content (i.e. same dynamic range and > frequency > response) encoded in both 24/96 and 16/44.1 formats, will flac > produce > roughly the same size compressed file for both formats?I'm not an expert here but I would guess probably not. the difference is in the noise, which a lossless codec has to preserve. if you do two recordings of the same noisy source, the one with the higher sample rate will have more information in the form of noise. a higher bit depth should theoretically not make as much of a difference assuming the DACs inject the same amount of noise, but I don't know if that holds in reality with higher bit depths. Josh __________________________________________________ Do You Yahoo!? Yahoo! - Official partner of 2002 FIFA World Cup http://fifaworldcup.yahoo.com