[mythtv] Profiling myth
Stuart Morgan
stuart at tase.co.uk
Mon May 31 14:56:32 UTC 2010
On Monday 31 May 2010 15:31:36 James Courtier-Dutton wrote:
> ... or are you just telling me not to even bother?
>
I'd welcome you to try and improve the speed, I just want to manage the
expectations. Every few months someone comes along with lots of enthusiasm for
improving it and every time the changes they end up making are slight or non-
existent because the vast array of hardware, configurations, broadcasters,
network configurations and more just make it an extremely difficult task.
Many of the buffers you are talking about didn't start off the size that they
are now, they've grown and grown in response to user experience. Adaptive
buffering was even added to try overcome this, but even the defaults for that
were arrived at over months/years of experimentation. Every time a few hundred
ms is shaved off it ends up being put back on again because it causes
regressions for someone.
MythTV is too frequently compared against a STB, but that's a fallacy. It's
very easy to tune things to that extent when you eliminate all the variables
such as hardware (including speed), broadcaster idiosyncrasies and network
latencies and reliability.
So please go ahead and we wish you the best of luck, but if you receive a
lukewarm response from the developers then you'll understand that it's because
we've been through this many, many times already. You might as well be telling
us that this time next year we'll all be living on the moon.
--
Stuart Morgan
More information about the mythtv-dev
mailing list