[mythtv] [PATCH] position map using frame numbers, rev 2

[Isaac Richards]

On Sunday 14 December 2003 12:03 am, Doug Larrick wrote:
> On 12/13/03 23:09:33, Isaac Richards wrote:
> > One more question, and I'll apply it, I promise =) For the single
> > frame advance (or any exact seek after the current keyframe and
> > before the next), does it seek to the nearest keyframe then decode up
> > to where it needs to be, or just decode up to where it needs to be?
> > Obviously, skipping the seek would be desired if possible.
>
> No, it does do the seek to the prior keyframe and then a decode of
> potentially ~1/2 sec. worth of frames.  Obviously, with rewind we have
> no choice since we can't decode backwards frame-by-frame.
>
> It's pretty simple to make the change you propose.  You can either
> apply what I've given you and I'll send a separate patch to do that, or
> you can wait for me to send in the whole mess one more time.  I may get
> a chance to look at it tomorrow (Sunday); if not, it will have to wait
> until Tuesday.

Started applying this, but PosMapFromEnc apparently doesn't work with the 
pvr-250..  Doesn't ever fill anything in.  I'll look into that more tomorrow.

Also, I'm not sure about that function -- it's getting called fairly often, 
and each call is going to be 3 separate queries to the backend (IsRecording, 
GetFramesWritten, and FillPositionMap).  The old code attempted to minimize 
talking to the backend like that, only querying it when it would help out, as 
in with a seek forward into unknown video data.  Doesn't really make much 
sense to ask the backend that often for information we already have or can 
figure out.

> BTW: what would you say to a patch that, when hdtvrecorder has no
> signal, generates & encodes a blue screen with text "No Signal" and
> sends that to the ringbuffer?  Might even be fancy and include the bad
> station's signal strength as reported by the card.  This would allow
> the user to exit in that situation w/o having to kill the frontend.

That'd be cool..  Could do the same with a message to the frontend, though, 
and that'd probably be a little easier than generating a mpeg stream.

Isaac