[mythtv-users] Long delays with remote occasionally

Josh Mastronarde jmastron at gmail.com
Sun Jun 21 06:01:48 UTC 2009


On Sat, Jun 20, 2009 at 2:36 AM, Josh Mastronarde<jmastron at gmail.com> wrote:
> On Sat, Jun 20, 2009 at 1:52 AM, Jean-Yves Avenard<jyavenard at gmail.com> wrote:
>> Hi
>>
>> 2009/6/20 Josh Mastronarde <jmastron at gmail.com>:
>>> Definitely let me know when your files are uploaded, and where I can
>>> download them from;  Yeechang sent me a file but it played fine for
>>> me; I'd like to reproduce a failure so I can look at that format
>>> decoding.
>>
>> Sorry, I thought I did.
>> http://www.avenard.org/files/media/videotest/
>>
>> badrecording.mp3 won't play with your patch
>
> Hrmmm, it plays for me (assuming you meant badrecording.mpg) on r20212

Okay, I'm busy wiping egg of my face now -- I just discovered that the
installation I ran both of your files on wasn't properly "make" or
"make install"ing somehow -- the /usr/local/lib files showed new
dates, but debug print statements I added weren't showing up until I
manually deleted .o and .so files and remade.  So I'm pretty sure that
when I said I could view both files, it really *didn't* have the patch
-- when I re-applied the patch I couldn't play them.

The good news is that I found trunk changeset 18965
(http://svn.mythtv.org/trac/changeset/18965) addresses this exact
issue that came up in trunk after the mpeg.c change was applied there
(it touches *other* files but not mpeg.c, which is why I didn't see it
before).  I applied the pieces of this in my now-fixed-setup, and I am
now able to play both of your "bad files".  This was all on my
original 20212 version; I will create a clean latest 0.21-fixes, make
a new backport patch, and confirm there.

The bad news is that this changeset affects 4 more files -- 2 of them
are in libavformat, which for 0.21-fixes hasn't changed in 13 months,
and 2 of them in libmythtv, which has many other recent 0.21-fixes
changes.

What does this mean?  For others following along:

- My original patch and binary libavformat.so are still good for
solving what seem to be the majority of mp4/mpegts detection issues,
and the long remote delays that seem to come from the misprobing,
provided that you don't have files (primarily older transcoded ones)
that fail to detect with the patch (you'll know if the it just jumps
back to the recordings screen).  This one can apply to a wide range of
0.21-fixes revs.

- I will try to make a binary libmythtv, but it will only be
applicable to the very latest 0.21-fixes (and may not apply to revs
released beyond this).  Only needed if you have the issues above

- I will make a complete source patch that may be applied to any
recent rev (as of now), and submit that as an addendum to my original
trac issue.  No guarantees whether the devs will apply to the
0.21-fixes stream (and I don't blame them if they don't want to apply
a 5-file patch)

Have I said how annoying the whole codec/file-format issue is, with
these insane "detection" schemes required to tell one from another?

Josh


More information about the mythtv-users mailing list