[mythtv] Ticket #2531: mythreplex recent regression in svn

Geoffrey Hausheer mythtv0368 at phracturedblue.com
Sun Oct 8 14:47:51 UTC 2006


On 10/7/06, D. Hugh Redelmeier wrote:
> I strongly recommend that regression tests be distributed.  It should
> be considered normal for all developers and even builders of Myth to
> run regression tests.

Here is why this isn't going to happen...
1) All of my test streams are copyrighted.  Some may be
redistributable if I could get permission (which requires that I know
where I got them, which I don't).  Also, the test-bench comes in at
near 1GB.
2) The only parts that would be easy to write regression tests for are
things like mythcommflag and mythtranscode.  The bulk of myth would be
much more difficult, and I really doubt any of the devs are going to
jump on such a monumental and tedious task
3) My regression stuff is very crude, quite incomplete, and often
broken (like at the moment).  I don't have the bandwidth to support
it.
4) It already requires a lot of very tedious effort to find and fix
transcoder bugs.  I'm not willing to spend a lot more effort
hand-crafting non-copyrighted test-cases.  That would just be a waster
of my time.

So, at least for the MPEG2-transcoder, things will continue to be as
they have been.  If you can show me a bug, I'll do my best to make
sure that stream continues to work with the transcoder in the future.


More information about the mythtv-dev mailing list