[mythtv] Google Summer of code 2006 - MythTV-related projects?

[Cymen Vig]

On 4/16/06, Ross Campbell <ross.campbell at gmail.com> wrote:
> Seeing the recent announcement for Google's summer of code 2006, I
> couldn't help but think that there are a few dozen if not hundreds of
> potential projects relating to MythTV that we could consider
> submitting...
>
> http://code.google.com/summerofcode.html
>
> We already have a good start on a lot of suggestions here:
> http://mythtv.org/wiki/index.php/Feature_Wishlist

Has anyone thought more about this? I am a college student working on
an undergraduate Computer Sciences degree at the University of
Wisconin at Madison. I'd love to participate in the Google Summer of
Code program and work on MythTV.

I have been tweaking my MythTV system with the various real time
options using Con Kolivas' kernel patches along with the schedtool and
ionice utilities (using the CFQ IO scheduler). With the options below,
live TV changes channels very quickly -- on the time of a digital set
top box.

The scheduling options:

High priority (real time):
ivtv-enc: schedtool -R -p 99, ionice -c1 -n0
mythfrontend: schedool -R -p 90, ionice -c1 -n7
mythbackend: schedtool -R -p 1, ionice -c1 -n3

Medium priority (similar to real time):
lircd, mysqld, apache2: schedtool -I, ionice -c2

Low priority (batch real time):
mythcommflag, mythtranscode: schedtool -B, ionice -c3

Kernel options:
Linux mythtv 2.6.15-ck7 #6 PREEMPT
CONFIG_PREEMPT=y
CONFIG_PREEMPT_BKL=y
CONFIG_IOSCHED_CFQ=y
CONFIG_DEFAULT_CFQ=y
CONFIG_DEFAULT_IOSCHED="cfq"

Hardware:
Cerelon-D 2.93 GHz, Hauppauge PVR-500

I still experience some IOBOUND issues when deleting files (on XFS
which is already tweaked to improve performance). File deletion is in
the backend process which I've got running at the "lowest of the
highest priority" class. If file deletion were in a separate process,
it would be simple to lower priority on it and I suspect that would
resolve the IOBOUND issue. Perhaps it would be far cleaner to add
scheduling capabilities to MythTV so things like recording scheduling,
file deletion, and other housework would run at a lower priority?

Integrating the use of priorities for processor and I/O resources into
MythTV would allow finer control of the various aspects of the system.
Some of the features I am using are not in the mainstream kernel yet
but in time, hopefully they will join Linus' tree. I suspect that this
work would not be a huge project but in combination with some of the
wish list items (referenced above) I think the project would benefit.

So any interest? I am interested from the applicant side and I'm sure
there are more students who would apply if the project signed on with
the Google Summer of Code.