[mythtv-users] What major features are planned for 0.27?
peper03 at yahoo.com
Wed Nov 21 09:58:20 UTC 2012
On 20/11/12 18:17, Mark Kendall wrote:
> On 19 November 2012 23:10, Richard <peper03 at yahoo.com> wrote:
>> On 19/11/12 23:12, Neil Salstrom wrote:
>> I've got one or two other areas of DVD playback I want to look at as I too
>> have various DVDs that don't play nicely/cleanly.
>> I can't give any promises or estimates as tracking down the causes is quite
>> time-consuming and fixing problems can be as simple as changing a couple of
>> lines (not a big deal) to having to change part of the architecture
>> (potentially a big deal).
>> I'm also not a myth-dev, just an unaffiliated-dev :) So that means I can
>> only create a patch and submit it with a ticket. It's up to one of the
>> myth-devs to find the time to review, (hopefully) accept and apply it.
> I'd suggest you revert:-
> before trying to fix anything. That commit will break playback for a
> significant minority of disks where the first title is a still frame
> or short menu and hence is a regression in 0.26 (and 0.25?). I think
> the actual memory leak was fixed subsequently and that element of code
> has nothing to do with 'guessing the main title' and everything to do
> with setting up the AVCodecContext correctly.
> FWIW, the dvd code is just terrible and I'd recommend tearing out
> every mention of 'dvd' in libmythtv and starting from scratch. It
> tries and fails to manage a three way state between the player, the
> ringbuffer and the decoder, for some reason uses the position map code
> and trying to manage the video buffers for still frames/menus etc for
> all the combinations of decoders is just painful. The latter is not
> helped by the fact that the player code was designed around video
> playback and is largely unsuited to handling the different use cases
> that dvd's throw at it.
Ok, I think you know more than enough about the inner workings of the
code so I'll readily accept your word. I'm not sure I'm in a position
to re-write everything, though. That seems to be a pretty major
undertaking even when you understand the overall Myth architecture,
which I don't yet.
I'm not sure how to continue at this point. Plug away at the existing
code, improving it where possible whilst gaining an understand of where
it's gone wrong architecturally, or jump straight in with a re-write
accepting the fact that I might need to re-write it more than once as I
realise where I'm going wrong architecturally? And then hope that my
work would be accepted :)
Obviously this can also be taken up by anyone else. So far, all I've
done is to submit a couple of minor bugfixes. By no means do I think
that this is all down to me.
More information about the mythtv-users