DebianSlug Arm package

[rpedde]

Ron,

Thanks.
My packages come from here:
http://ipkg.nslu2-linux.org/feeds/optwa ... ss/stable/
I thought that unslung stood for optware.

Well, I recompiled, but got the same error with the patched vorbis-tools:

Code: Select all

checking for Vorbis... no
*** Could not run Vorbis test program, checking why...
*** The test program failed to compile or link. See the file config.log for the
*** exact error that occured. This usually means Vorbis was incorrectly installed
*** or that you have moved Vorbis since it was installed.
configure: error: Vorbis needed!


In config.log:

Code: Select all

/opt/lib/gcc/mipsel-linux-uclibc/4.1.1/../../../libvorbis.so: undefined reference to `lrint'


Essentially the same as yesterday.

I don't expect you to come with a solution though. I hope the ivorbis-tools package will be repaired soon.

Marc

Hrm. I didn't use the oleg chain, I was using th wl-500g build chain. I'll try oleg probably later this weekend.

-- Ron

[mumsoft]

Ron,

Hrm. I didn't use the oleg chain, I was using th wl-500g build chain. I'll try oleg probably later this weekend.

I would appreciate that. Meanwhile I have asked Sollie (he has compiled mt-daapd on his WL-700gE too) to repeat my steps. It could be that my setup is not correct.

I don't know for sure, but I don't think that packages for the wl-500g are useful for the WL-700gE with KCFurge template. I might have tried some in the past and did not succeed.

[rpedde]

I don't know for sure, but I don't think that packages for the wl-500g are useful for the WL-700gE with KCFurge template. I might have tried some in the past and did not succeed.

That could be, but the unslung build chain is really featureful, and easy to repeatably cross-compile. Doing random cross-compilation without the unslung build-harness is fraught with problems, in my experience.

Every time I've tried to port to something with cross-compiler, it was always hit or miss -- some packages compiled with different options, etc.

If the wl-700ge firmware could be made compatible with unslung, it would be a win for everyone. Not the least of all, users, who would have thousands of packages to choose from.

I guess it's easy to see I'm something of an unslung bigot -- seems like it's easy to get a platform working with unslung, and once that's done, the benefits are huge.

Meanwhile, I didn't yet compile with oleg. I'll still try it. If nothing else, you can untar the ipkg and see if it runs.

-- Ron

[sollie]

Can somebody tell me wich command to use, for adding the patch.

Sollie.

[mumsoft]

Hi Sollie,

Can somebody tell me wich command to use, for adding the patch.

Sollie.

I thought I taught you that...

Code: Select all

patch -p1 <oggdec-tremor.patch

Do this in the directory where you have the directory

Code: Select all

vorbis-tools-1.0

and both patch-files of course.

This gives me this 'error':

Code: Select all

can't find file to patch at input line 4
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|diff -ur vorbis-tools-1.0.dist/ogg123/Makefile.in vorbis-tools-1.0/ogg123/Makefile.in
|--- vorbis-tools-1.0.dist/ogg123/Makefile.in   2002-07-19 09:19:14.000000000 -0500
|+++ vorbis-tools-1.0/ogg123/Makefile.in        2005-04-13 01:23:38.000000000 -0500
--------------------------
File to patch:

Then enter (cut & paste)

Code: Select all

vorbis-tools-1.0/ogg123/Makefile.in

This repeats several times with different files it can't find, due to the difference in the date. And that does not matter here.

Succes
Marc

[sollie]

Mumsoft,

I got exactly the same problems you got. And i dont know what todo.

Sollie.

[mumsoft]

Hi Sollie,

Mumsoft,

I got exactly the same problems you got. And i dont know what todo.

Sollie.

Thanx for trying in the first place.
But I did not expect this. Why? Well, because today I came over the vorbis error, and my heart jumped some meters. But now it stops with

Code: Select all

../libtool: line 1309: cd: yes/lib: No such file or directory
libtool: link: cannot determine absolute directory name of `yes/lib'
make[2]: *** [oggenc] Error 1
make[2]: Leaving directory `/shares/MYVOLUME1/MYSHARE1/firefly/vorbis-tools-1.0/oggenc'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/shares/MYVOLUME1/MYSHARE1/firefly/vorbis-tools-1.0/oggenc'
make: *** [all-recursive] Error 1


To come so far, I have reinstalled uclibc-opt, while originally I had installed uclibc. There are 2 packages in the oleg branch, one is called uclibc-opt, the other uclibc. I thought that I needed uclibc, not uclibc-opt, but that is a mistake.

Now I hope Ron can give me a clue about this new error.

[mumsoft]

To come so far, I have reinstalled uclibc-opt, while originally I had installed uclibc. There are 2 packages in the oleg branch, one is called uclibc-opt, the other uclibc. I thought that I needed uclibc, not uclibc-opt, but that is a mistake.

Now I hope Ron can give me a clue about this new error.

For the first time I own a Roku Soundbridge, using mt-daapd as streaming server with ogg as my preferred music format, I can actually hear the Music. "Communication is the problem to the answer", sings 10CC.

Ron, with the above mentioned quircks I could configure (i)vorbis-tools(1.0) without errors. But, as said, make gave errors about yes/lib not being there, so I edited several makefiles (in oggdec and oggenc etc) to change yes/lib into /opt/lib. I had configured with --prefix=/opt, but apparently that did not lead to sound makefiles. Now Make proceeded a bit further, but then gave another error in ogg123.
I got mad...
Now I edited the Makefile in ivorbis-tools, and removed every subdir except oggdec. I ran make clean. I ran make again... no errors(!). I ran ipkg install ivorbis-tools. I ran make install. This installed ONLY oggdec, BUT still the big endian, noise producing version! (? what about your patch?)
I got madder. Angry. So I edited oggdec.c. Removed the big-endian stuff. Ran make clean, make, make install... No, I copied the resulting oggdec binary.
And now finally, finally, at last, this damn Roku does not produce noise anymore.

I was planning a How-to, but this is just to laborious to reproduce. So I will offer my precious little oggdec for download, if possible.

Still, I wonder: what on earth did Xaviour that he could come away with, ha, it works, thanx, goodbye, good luck...

[thorstenhirsch]

Alright, today I updated the tremor patch so that it can be applied to libvorbis-1.1.1. This patch also includes the endian patch. Here it is:

Code: Select all

--- ogg123/Makefile.in~ 2005-06-27 11:29:11.000000000 +0200
+++ ogg123/Makefile.in  2007-10-07 22:05:41.000000000 +0200
@@ -209,7 +209,7 @@
USE_NLS = @USE_NLS@
VERSION = @VERSION@
VORBISENC_LIBS = @VORBISENC_LIBS@
-VORBISFILE_LIBS = @VORBISFILE_LIBS@
+VORBISFILE_LIBS = -lvorbisidec
VORBIS_CFLAGS = @VORBIS_CFLAGS@
VORBIS_LIBS = @VORBIS_LIBS@
XGETTEXT = @XGETTEXT@
@@ -270,7 +270,7 @@
            @PTHREAD_CFLAGS@ @SHARE_CFLAGS@ @I18N_CFLAGS@

ogg123_LDADD = @SHARE_LIBS@ \
-              @VORBISFILE_LIBS@ @VORBIS_LIBS@ @OGG_LIBS@ @AO_LIBS@ \
+              -lvorbisidec @OGG_LIBS@ @AO_LIBS@ \
                @SOCKET_LIBS@ @LIBICONV@ @CURL_LIBS@ @PTHREAD_CFLAGS@ \
                @PTHREAD_LIBS@ @I18N_LIBS@ @FLAC_LIBS@ @SPEEX_LIBS@

--- ogg123/cmdline_options.c~   2005-06-13 15:11:44.000000000 +0200
+++ ogg123/cmdline_options.c    2007-10-07 22:05:34.000000000 +0200
@@ -195,7 +195,7 @@
        break;

       case 'V':
-       status_error(_("ogg123 from %s %s\n"), PACKAGE, VERSION);
+       status_error(_("ogg123 (w/Tremor hack) from %s %s\n"), PACKAGE, VERSION);
        exit(0);
        break;

--- ogg123/oggvorbis_format.c~  2005-06-03 12:15:09.000000000 +0200
+++ ogg123/oggvorbis_format.c   2007-10-07 22:05:28.000000000 +0200
@@ -21,7 +21,7 @@
#include <ctype.h>
#include <ogg/ogg.h>
#include <vorbis/codec.h>
-#include <vorbis/vorbisfile.h>
+#include <tremor/ivorbisfile.h>
#include "transport.h"
#include "format.h"
#include "vorbis_comments.h"
@@ -136,9 +136,7 @@
   while (nbytes > 0) {

     old_section = priv->current_section;
-    ret = ov_read(&priv->vf, ptr, nbytes, audio_fmt->big_endian,
-                 audio_fmt->word_size, audio_fmt->signed_sample,
-                 &priv->current_section);
+    ret = ov_read(&priv->vf, ptr, nbytes, &priv->current_section);

     if (ret == 0) {

--- oggdec/Makefile.in~ 2005-06-27 11:29:12.000000000 +0200
+++ oggdec/Makefile.in  2007-10-07 22:08:24.000000000 +0200
@@ -183,7 +183,7 @@
USE_NLS = @USE_NLS@
VERSION = @VERSION@
VORBISENC_LIBS = @VORBISENC_LIBS@
-VORBISFILE_LIBS = @VORBISFILE_LIBS@
+VORBISFILE_LIBS = -lvorbisidec
VORBIS_CFLAGS = @VORBIS_CFLAGS@
VORBIS_LIBS = @VORBIS_LIBS@
XGETTEXT = @XGETTEXT@
--- oggdec/oggdec.c~   2005-06-03 12:15:08.000000000 +0200
+++ oggdec/oggdec.c    2007-10-08 00:06:05.000000000 +0200
@@ -12,27 +12,25 @@
#include <getopt.h>
#include <errno.h>
#include <string.h>
+#include <endian.h>

#if defined(_WIN32) || defined(__EMX__) || defined(__WATCOMC__)
#include <fcntl.h>
#include <io.h>
#endif

-#include <vorbis/vorbisfile.h>
+#include <tremor/ivorbisfile.h>

struct option long_options[] = {
     {"quiet", 0,0,'Q'},
     {"help",0,0,'h'},
     {"version", 0, 0, 'v'},
-    {"bits", 1, 0, 'b'},
-    {"endianness", 1, 0, 'e'},
     {"raw", 0, 0, 'R'},
-    {"sign", 1, 0, 's'},
     {"output", 1, 0, 'o'},
     {NULL,0,0,0}
};

-#define VERSIONSTRING "OggDec 1.0.1\n"
+#define VERSIONSTRING "OggDec 1.0.1 (Tremor)\n"

static int quiet = 0;
static int bits = 16;
@@ -49,11 +47,6 @@
                     " --quiet, -Q      Quiet mode. No console output.\n"
                     " --help,  -h      Produce this help message.\n"
                     " --version, -v    Print out version number.\n"
-                    " --bits, -b       Bit depth for output (8 and 16 supported)\n"
-                    " --endianness, -e Output endianness for 16 bit output. 0 for\n"
-                    "                  little endian (default), 1 for big endian\n"
-                    " --sign, -s       Sign for output PCM, 0 for unsigned, 1 for\n"
-                    "                  signed (default 1)\n"
                     " --raw, -R        Raw (headerless) output.\n"
                     " --output, -o     Output to given filename. May only be used\n"
                     "                  if there is only one input file\n"
@@ -68,7 +61,7 @@
     int option_index = 1;
     int ret;

-    while((ret = getopt_long(argc, argv, "Qhvb:e:Rs:o:",
+    while((ret = getopt_long(argc, argv, "QhvR:o:",
                     long_options, &option_index)) != -1)
     {
         switch(ret)
@@ -84,19 +77,6 @@
                 fprintf(stderr, VERSIONSTRING);
                 exit(0);
                 break;
-            case 's':
-                sign = atoi(optarg);
-                break;
-            case 'b':
-                bits = atoi(optarg);
-                if(bits <= 8)
-                    bits = 8;
-                else
-                    bits = 16;
-                break;
-            case 'e':
-                endian = atoi(optarg);
-                break;
             case 'o':
                 outfilename = strdup(optarg);
                 break;
@@ -242,7 +222,7 @@
         }
     }

-    while((ret = ov_read(&vf, buf, buflen, endian, bits/8, sign, &bs)) != 0) {
+    while((ret = ov_read(&vf, buf, buflen, &bs)) != 0) {
         if(bs != 0) {
             fprintf(stderr, "Only one logical bitstream currently supported\n");
             break;
@@ -255,6 +235,18 @@
             continue;
         }

+#ifdef BIG_ENDIAN
+        { /* byteswap */
+            int i;
+            char tmp_char;
+            for(i=0;i<ret;i+=2) {
+                tmp_char = buf[i];
+                buf[i] = buf[i+1];
+                buf[i+1] = tmp_char;
+            }
+        }
+#endif
+
         if(fwrite(buf, 1, ret, out) != ret) {
             fprintf(stderr, "Error writing to file: %s\n", strerror(errno));
             ov_clear(&vf);

However, 1 hunk will fail for oggdec.c. I don't know why, I've created a new diff 3x with "diff -ruN", but somehow this one damn line can't be patched correctly. Maybe it's also too late and I'm too tired.

The next problem: I still can't play ogg files.
This is what I did:

1.) download tremor (svn co http://svn.xiph.org/trunk/Tremor/ tremor)
2.) autogen, configure, make, make install on the LS2
3.) download vorbis-tools-1.1.1
4.) apply my patch (and change this 1 line manually)
5.) configure, make, make install

Everything run successfully and I can also run the decoding script of firefly manually w/o problems:

Code: Select all

/usr/local/bin/mt-daapd-ssc.sh "/som/ogg/file.ogg" 0 251.000 "ogg"

My LS2 tries to play the ogg file, but cancels at once and says, that it can't play the song. This is the log entry of firefly:

Code: Select all

2007-10-08 00:38:25 (00080003): Session 0: Streaming file 'someoggfile.ogg' to 192.168.0.22 (offset 0)

So no error and also no sound. The SB doesn't even say, that it waits for the song to be cached (that's what it did with the "normal" liboggdec/vorbis-tools).

[rpedde]

Everything run successfully and I can also run the decoding script of firefly manually w/o problems:

Code: Select all

/usr/local/bin/mt-daapd-ssc.sh "/som/ogg/file.ogg" 0 251.000 "ogg"

My LS2 tries to play the ogg file, but cancels at once and says, that it can't play the song. This is the log entry of firefly:

Code: Select all

2007-10-08 00:38:25 (00080003): Session 0: Streaming file 'someoggfile.ogg' to 192.168.0.22 (offset 0)

So no error and also no sound. The SB doesn't even say, that it waits for the song to be cached (that's what it did with the "normal" liboggdec/vorbis-tools).

Did you give a full path to oggdec in /usr/local/bin/mt-daapd-ssc.sh? As well as wavstreamer? Might be that they are on your path in your environment, but not in the environment that mt-daapd runs as.

-- Ron

[thorstenhirsch]

Hi Ron,

now I added the full path into mt-daapd-ssc.sh, but it still does not work. It's a little bit better now, because the SB doesn't say, that it can't play the song at once. Instead it says "preparing to play" for 7 seconds and then it says, that it can't play the song.

This is the process tress:

Code: Select all

  263 ?        S      0:00 /usr/local/sbin/mt-daapd start
  264 ?        S      0:08 /usr/local/sbin/mt-daapd start
  303 ?        S      0:00      |   \_ /bin/sh /usr/local/bin/mt-daapd-ssc.sh /mnt/hda/share/musik/niagara/la_verite/aux_coeurs_ble
  307 ?        R      1:00      |       \_ /usr/local/bin/oggdec --quiet -o - /mnt/hda/share/musik/niagara/la_verite/aux_coeurs_ble
  308 ?        S      0:00      |       \_ /usr/local/bin/wavstreamer -o 0 -l 251.000

So mt-daapd seems to try it at least. Is there supposed to be some STDOUT when calling mt-daapd-ssc.sh manually? Because there's no output at all. It also doesn't return to the shell, it just seems to do nothing.

Here's what strace tells me about these processes.
303 - mt-daapd-ssc.sh:

Code: Select all

$ strace -p 303
Process 303 attached - interrupt to quit
waitpid(-1,  <unfinished ...>
Process 303 detached

307 - oggdec
$ strace -p 307
Process 307 attached - interrupt to quit
Process 307 detached[/CODE]
308 - wavstreamer

Code: Select all

$ strace -p 308
Process 308 attached - interrupt to quit
read(0,  <unfinished ...>
Process 308 detached

So actually oggdec is doing nothing, but eats all my cpu power.

edit: without the --quiet parameter, oggdec tells me the following line a million times:

Code: Select all

Warning: hole in data

edit2: this means, that ov_read gives a bad return code. ov_read is defined in tremor's ivorbisfile.h/vorbisfile.c; maybe my tremor package is buggy - this is the version string in vorbisfile.c:

Code: Select all

$Id: vorbisfile.c,v 1.6 2003/03/30 23:40:56

[mas]

I have to say that whole oggdec-tremor thing is nasty. When I did it back then a while ago I also had to try around a lot and have since then deliberately not touched this patched oggdec thing any more.

Maybe it would be a simplification to move a small patched oggdec into the mt-daapd source and have it compiled/installed when a --with-oggdectremor option is given at the ./confogure call?

[rpedde]

So mt-daapd seems to try it at least. Is there supposed to be some STDOUT when calling mt-daapd-ssc.sh manually? Because there's no output at all. It also doesn't return to the shell, it just seems to do nothing.

Ya, it should dump to stdout.

[thorstenhirsch]

And what do you think about mas' suggestion? Can you include tremor's liboggidec and the patched oggdec into firefly? The license is BSD, so the legal aspect should be no problem at all.

[rpedde]

And what do you think about mas' suggestion? Can you include tremor's liboggidec and the patched oggdec into firefly? The license is BSD, so the legal aspect should be no problem at all.

I kind of hesitate to, as that's just more to maintain, plus it's already in all the ipkg formats.

I guess the only thing missing is debian packages. I'd think about doing debian packages. What other formats would be required?