Re: [Musicpd-dev-team] mpd uclinux branch is not available

2009-08-07 Thread Barry Song
Eric,
Are you available?
Thanks
Barry

On Fri, Aug 7, 2009 at 1:41 PM, Max Kellermannm...@duempel.org wrote:
 On 2009/08/07 06:54, Barry Song 21cn...@gmail.com wrote:
 As said in mpd website, there is a uclinux branch for it. But the
 branch of mpd is not available by the following address:
 http://musicpd.org/~normalperson/mpd-uclinux/
 http://mpd.bogomips.org/mpd-uclinux/
 git://mpd.bogomips.org/mpd-uclinux.git/
 http://mpd.bogomips.org/mpd-uclinux/mpd-uclinux.git/
 http://musicpd.org/~normalperson/mpd-uclinux/mpd-uclinux.git/

 Eric Wong (aka normalperson) used to work on that code, but I have
 never seen it myself.  Eric left the project one year ago.

 We are using mpd release 0.13.2 to run on no-mmu platform, there are
 many problems in it. There are 3 processes running at the same time
 fulfilled by fork, I tried to replace them by pthread to support
 multi-thread, then I still encountered many other problems.  After
 checking the 0.15.1 release, it seems you have deleted the fork and
 the multi-thread is now fulfilled by glib thread but not pthread, so
 it is not useful to me yet since we are using uclibc.

 What's the problem with GLib threading?  I have never worked on
 MMU-less architectures..

 Max


--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team


Re: [Musicpd-dev-team] mpd uclinux branch is not available

2009-08-07 Thread Barry Song
On Fri, Aug 7, 2009 at 1:41 PM, Max Kellermannm...@duempel.org wrote:
 On 2009/08/07 06:54, Barry Song 21cn...@gmail.com wrote:
 As said in mpd website, there is a uclinux branch for it. But the
 branch of mpd is not available by the following address:
 http://musicpd.org/~normalperson/mpd-uclinux/
 http://mpd.bogomips.org/mpd-uclinux/
 git://mpd.bogomips.org/mpd-uclinux.git/
 http://mpd.bogomips.org/mpd-uclinux/mpd-uclinux.git/
 http://musicpd.org/~normalperson/mpd-uclinux/mpd-uclinux.git/

 Eric Wong (aka normalperson) used to work on that code, but I have
 never seen it myself.  Eric left the project one year ago.

 We are using mpd release 0.13.2 to run on no-mmu platform, there are
 many problems in it. There are 3 processes running at the same time
 fulfilled by fork, I tried to replace them by pthread to support
 multi-thread, then I still encountered many other problems.  After
 checking the 0.15.1 release, it seems you have deleted the fork and
 the multi-thread is now fulfilled by glib thread but not pthread, so
 it is not useful to me yet since we are using uclibc.

 What's the problem with GLib threading?  I have never worked on
 MMU-less architectures..

If the system doesn't use GLib, for example uclibc or some other libc,
it can't use the g_thread. Maybe posix thread is a better choice?

 Max


--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team


Re: [Musicpd-dev-team] mpd uclinux branch is not available

2009-08-07 Thread Max Kellermann
On 2009/08/07 08:40, Barry Song 21cn...@gmail.com wrote:
 If the system doesn't use GLib, for example uclibc or some other libc,
 it can't use the g_thread. Maybe posix thread is a better choice?

Why does the system care about a userspace library?  gthread is just a
portability wrapper around pthread.

--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team


Re: [Musicpd-dev-team] mpd uclinux branch is not available

2009-08-07 Thread Max Kellermann
On 2009/08/07 09:15, Barry Song 21cn...@gmail.com wrote:
 Some embedded systems don't like to use glibc, but the smaller C
 library, for example uclibc. Then g_thread_init,  g_thread_create,
 g_thread_create and so on will not be available.

GLib and glibc are two completely unrelated projects.

--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team


Re: [Musicpd-dev-team] Fwd: hively decoder patch

2009-08-07 Thread Tony Miller
The author has replied to me in this thread:
http://hivelytracker.com/forum.php?action=viewthreadid=153

He simply says Credit is fine.

Apparently the issue has come up before in this thread:
http://hivelytracker.com/forum.php?action=viewthreadid=114

In this thread the author states The freely available replay code is public
domain. The replay code is what I used.


-Tony

On Tue, Jul 28, 2009 at 11:23 AM, Tony Miller mcfiredr...@gmail.com wrote:

 I posted the question to them on their forum.

 That is indeed where the code came from. Peter does appear to be the author
 of hively tracker. If the forums do not respond soon I'll just email him
 directly.

 There are more tunes here:
 http://amigascne.org/abyss/ahx/tunes.html

 -Tony


 On Tue, Jul 28, 2009 at 8:44 AM, Jeffrey Middleton jefr...@gmail.comwrote:

 Max, It looks like there are some samples available from the tunes
 section of the website - http://hivelytracker.com/tunes/

 Tony, I'm guessing the source code you used came from here:
 http://www.petergordon.org.uk/files/replayer_linux.tgz
 This is linked from the download section of the website, where it says the
 port is by Christopher O'Neill, but the readme says (and hosting suggests)
 it's by Peter Gordon (p...@petergordon.org.uk).  There's also a forum you
 could post in to ask.  It really does look like none of these people ever
 bothered to think about applying a license to their code, but given the
 number of ports and informality of the website it seems like they might be
 amenable to a GPL so it could go into MPD... good luck.

 Jeffrey

 On Tue, Jul 28, 2009 at 10:31 AM, Max Kellermann m...@duempel.org wrote:

 On 2009/07/25 19:32, Tony Miller mcfiredr...@gmail.com wrote:
  Subject: [PATCH] This is my patch for my hively/ahx decoder plugin.
 This
  allows you to playback .ahx and .hvl files. There is no library for
 this
  format, just hvl_replay.c and hvl_replay.h which are distributed by
  hivelytracker.com. This code does not appear to have any license. If
 this is
  merged, it should be a configure option, something like,
  --enable-hively.

 If the code does not have any license, who allowed you to copy it
 then?  Without explicit permission (usually given by the license),
 we're not allowed to use it.

 Please tell me where to download sample files in that format.  I've
 never heard of it, but I'm always happy to get new decoders into MPD
 (once the copyright issue is solved).

 Max


 --
 Let Crystal Reports handle the reporting - Free Crystal Reports 2008
 30-Day
 trial. Simplify your report design, integration and deployment - and
 focus on
 what you do best, core application coding. Discover what's new with
 Crystal Reports now.  http://p.sf.net/sfu/bobj-july
 ___
 Musicpd-dev-team mailing list
 Musicpd-dev-team@lists.sourceforge.net
 https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team




--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team


Re: [Musicpd-dev-team] mpd uclinux branch is not available

2009-08-07 Thread Eric Wong
Barry Song 21cn...@gmail.com wrote:
 Eric,
 Are you available?
 Thanks
 Barry

Hi Barry,

Check with Courtney Cavin (courtc) or Alastair Stuart (coob) with
the iPodLinux project.  I'm not sure where my backups are; a bunch of
stuff got lost/misplaced last year in the move.

 On Fri, Aug 7, 2009 at 1:41 PM, Max Kellermannm...@duempel.org wrote:
  On 2009/08/07 06:54, Barry Song 21cn...@gmail.com wrote:
  We are using mpd release 0.13.2 to run on no-mmu platform, there are
  many problems in it. There are 3 processes running at the same time
  fulfilled by fork, I tried to replace them by pthread to support
  multi-thread, then I still encountered many other problems.  After
  checking the 0.15.1 release, it seems you have deleted the fork and
  the multi-thread is now fulfilled by glib thread but not pthread, so
  it is not useful to me yet since we are using uclibc.
 
  What's the problem with GLib threading?  I have never worked on
  MMU-less architectures..

pthreads didn't work well at the time in my experience, I had to use
clone() (this was back in the day with uclinux 2.4):

  - LinuxThreads always spawns an extra manager thread in the background

  - clone() allows per-task signal handlers, so it was easier to keep
the old signal handler code from 0.1[123].x

  - 2.4 also didn't have posix_fadvise, so I needed an extra thread to
do background read ahead (no mmap + madvise since there
was no MMU, either).

  - pthreads enforces a minimum stack size of 64K (the default is
2M(!)), I largest stack I ever needed was much smaller than that and
I had readahead thread using ~500 bytes.  malloc is (or was at the
time) very wasteful in uclinux.

All that code was based around the limitations of the uclinux 2.4 kernel
at the time, so if you're using 2.6 it may not be worth it...

--
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
___
Musicpd-dev-team mailing list
Musicpd-dev-team@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/musicpd-dev-team