xmms-musepack crashes on amd64 bit machine
 

Has anyone else experienced this? Seems to crash it every time I try to play an mpc file. I'll look into this if nobody else is already.

Ok. I think I've figured out what's going on here.

The current xmms-musepack plugin code links in both glib-1.2 and glib-2.0. According to this bug in the redhat bug database:

https://bugzilla.redhat.com/bugzilla/show_b...g.cgi?id=131609

this is a no-no. I'm not sure why it only manifests itself on amd64 machines.

As a workaround I've just commented out the calls to the glib-2.0 utf8 string conversions functions, which seem to be the only reason the code needs glib-2.0. As an added bonus, I've ported the xmms plugin code to the new C libmusepack.

This work in progress is available on my svn server (http://www.caddr.com/svn and http://www.caddr.com/websvn).

It seems like a better solution would be to change the code to use libiconv for character conversion, right? At least I can listen to my mpc files on my amd64 machine now. :)

And then I lose all my unicode tags, so the libiconv is a better solution indeed.

Ok. I've switched the conversion routine over to iconv and checked in the patch in svn. Seems to be working fine for reading tags, but I can't seem to write tags using the XMMS plugin.

Was this working before? I don't think I ever tried it.

I've ported the bmp plugin over to the new C libmusepack as well. All the code is in my svn, as usual.

Also, here's the glib -> iconv patch for the current xmms-musepack-1.1, if we want to release an updated version without waiting for the new C libmusepack.

diff -ru xmms-musepack-1.1/src/libmpc.cpp xmms-musepack-1.1-patched/src/libmpc.cpp
--- xmms-musepack-1.1/src/libmpc.cpp 2004-11-28 08:15:57.000000000 -0800
+++ xmms-musepack-1.1-patched/src/libmpc.cpp 2004-12-29 15:21:10.810832296 -0800
@@ -12,6 +12,7 @@
}
#include <glib.h>
#include <gtk/gtk.h>
+#include <iconv.h>
#include <math.h>
#include "tags.h"
#include "equalizer.h"
@@ -115,12 +116,28 @@
}
#endif

-static char* convertUTF8toLocale(char* utf8) {
- char* temp=g_locale_from_utf8 (utf8, -1, NULL, NULL, NULL);
- if(temp==NULL)
- return g_strdup(utf8);
- else
- return temp;
+static char*
+convertUTF8toLocale(char* utf8)
+{
+ // note - opens a new iconv descriptor for each call
+ // will have to find a way to reuse the descriptor if this turns
+ // out to be too slow
+ iconv_t idesc = iconv_open("", "UTF-8");
+ if (idesc == (iconv_t) -1) {
+ perror("iconv_open failed");
+ return g_strdup(utf8);
+ }
+
+ size_t in_left = strlen(utf8);
+ size_t out_left = 2 * in_left + 1;
+ char *buf = (char *)g_malloc(out_left);
+ char *in = utf8;
+ char *out = buf;
+
+ memset(buf, 0, out_left);
+ size_t err = iconv(idesc, &in, &in_left, &out, &out_left);
+ iconv_close(idesc);
+ return buf;
}

static void convertLE32to16(MPC_SAMPLE_FORMAT* sample_buffer, char* xmms_buffer, unsigned int status) {

Updated !

INSTALL still says:
and in README theres no mention of changes for 1.1.1

Shouldn't Glib-2.0 be removed from list of dependencies? And aren't iconv a new dependency - or did I get that wrong?

Forgot to update text files... Fixed package will be online soon. Code is unchanged.

I've ported the lamip plugin over to libmusepack as well. It's in my svn.

i've told fred (lamip dev) of your changes and he'll port it to the 0.0.3 API

thanx kuniklo :)

the lamip devs could use some help if you are interested in the player


later

Great. I'll see if I can find any time to help out with lamip. I've got several things I want to do with musepack first.