[mythtv] tiny code improvement.

[Isaac Richards]

On Sunday 11 May 2003 06:10 pm, michael at optusnet.com.au wrote:
> Matt Zimmerman <mdz at debian.org> writes:
> > On Sun, May 11, 2003 at 05:19:57PM +1000, michael at optusnet.com.au wrote:
> > > #2. Mythtv is very slow to change position. Skipping forward
> > > 30sec or 10 min can take 10's of seconds.
> >
> > Not here.  Have you investigated the root cause?  This is a good first
> > step when trying to solve a problem. :-)
>
> Trying to. I leave vmstat running in a window and hit page-up. CPU
> utilization falls to zero, disk activity stops, mythtv freezes for
> 20 seconds and then continues. Want to take a stab at where you
> think the problem lies? :)

If you've been reading the mailing list, you'dve seen that remote seeking on 
live-tv or a still-recording file is slower than a recorded file that's 
finished recording if it needs to request multiple keyframe positions during 
a single seek, as the current code doesn't attempt to aggregate the requests 
into a single one.  Bruce Markey's promised an enhancement to fix this soon 
=)

Isaac