[mythtv] MythRoku - trying to mimic mythfrontend frame seeking

[Daniel Walton]

On Wed, 1 Feb 2006, Geoffrey Hausheer wrote:

> On 2/1/06, Daniel Walton wrote:
> > So when the frontend begins to play a recording it queries the SQL database for
> > the frame<->offset info?  I imagine it also queries for the commercial skip info
> > at this time too.  Are there any documents of the syntax of these queries?
> Yes.  the mythfrontend does something like:
> select mark,offset from recordedmarkup where chanid=xxx and
> starttime=yyy and type=zzz
>
> type can be:
> 7: keyframe (mark is based on the keyframe number)
> 9: byframe (mark is based on the actual frame number)
> 4: commercial start
> 5: comeercial end
>
> You also need to get the cutlist via:
> select cutlist from recorded where chanid = xxx and starttime = yyy
>
> As I said though, this is likely notthe ideal way to do things for the
> photobridge, especially because if type '7' was used, it'll have a
> hard time determining where it is.  Of course after 0.19 is released,
> type '7' will likely go away, which will make this much easier. to
> deal with.

Thanks for all the info.  Let's assume type 7 goes away, why couldn't the
photobridge use type 9, just because of the limited memory on the photobridge?
Roughly how much memory does it take a hold the position map for an hour long
show?

My understanding is that most folks with a Rev A photobridge (64M of mem)  are
already using a swap file on a compact flash card to run third party apps.  If
adding 64M or 128M of swap space gives the ability to do commercial skip then
that seems like a reasonable tradeoff to me :)

Daniel

>
> For reference:  type 7 was orginally designed for recoridng with a
> PVR250 (which is pretty reliable at keeping a fixed distance between
> gops) but it can't be used for DVB or ATSC (or transcoded material),
> and it isn't reliably applied on some of these types of material (for
> these we use type 9).  Plus it makes jumps more complicated because
> myth needs to keep track of 2 different methods of doing the same
> thing.  In a desire to make flagging/cutting more accurate, we will
> likely try to remove the type 7 stuff.
>
> .Geoff
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
>