[mythtv] Re:Channel change speed

[Ian Forde]

Agreed.  Even with a DCT2000 cable box and the serial scripts, my
script[1] on its own takes ~3.5 seconds to complete a channel change. 
Add the ringbuffer stuff, and now I'm in the 5-6 second range.  Before
hearing about the ringbuffer stuff, I tried to speed up my script. 
Wasn't much I could do there unless I wanted to get really into some
really ugly stuff.

I wonder if it might be better to have the channel change and ringbuffer
teardown run simultaneously... Couldn't they be separate threads?  Or at
least if an external script is used, can't it be run without waiting for
it to finish, have the ringbuffer stuff take place, then check back for
the results of the script?

	-I

[1] - The script has been updated.  You can get them at
http://jeigh.com/~ian/dct2000serial.tar.gz

On Thu, 2003-09-11 at 20:27, Matt Zimmerman wrote:
> On Thu, Sep 11, 2003 at 04:44:08PM +0100, matt.jarvis at philips.com wrote:
> 
> > Search the list archives, there is a C program for channel changing with
> > an external tuner which is WAY faster than the scripts, and was posted to
> > the list. I've always wondered why it's not in the myth contrib
> > directory....
> 
> It's a bit silly to write the thing in C when it's not the bottleneck.  The
> only significant delay in the script is where it explicitly waits, to give
> it a better chance of the signal being recognized.  Just decrease the sleep
> time if it's still reliable for you.
-- 
                       __________________________________
                       Ian Forde, RHCE, CCSE, SCNA, SCDME
                       CYTBeN, Inc.
                       ian at duckland.org / ian at cytben.com