[mythtv-users] How does (should) mythtv match repeats?

[Mike Perkins]

Chris Simmons wrote:
> Hi all,
> 
> I've noticed recently that myth has sometimes been recording repreats of 
> shows.  I use UK EIT information for my scheduling data.  I'm running 
> atrpms 0.20.2-172.fc8.
> 
> To take an example, I've set up myth to record "Torchwood" any time on 
> BBC2.  But it recorded a showing on Wednesday, with description
> 
> "Rhys discovers the truth about Torchwood and becomes part of the team 
> as they investigate a mysterious alien meat supply. Some strong 
> language. 4/13."
> 
> and now it wants to record a showing thats obviously (to a human) a repeat:-
> 
> "Rhys discovers the truth about Torchwood and becomes part of the team 
> as they investigate a mysterious alien meat supply. 4/13."
> 
> As you can see though the description isn't an exact match.  Presumably 
> myth decides shows are repeats only if the description is an exact 
> match?  I've also seen other minor differences in repeats like 
> corrections for typos, removing a comma etc and ended up having to 
> delete these recordings.
> 
> So my questions are
> 
> 1) Does myth require an exact match?
> 
> 2) Is this reasonable - wouldn't it be better if myth used approximate 
> text matching with some kind of threshold or something?
> 
> 3) Is there anything like this on the way for the .21 release?
> 
This seems to happen quite a bit recently, even with the xmltv-supplied data. I 
get it on Torchwood, also on Primeval, Sky at Night, etc. Since the bulk of the 
description must have been cut'n'pasted, why on earth would anyone then go in 
and subtly alter just *one* of the duplicates? Is this a question some should be 
asking of Radio Times (who provide the data)?

I note that for some of the programs affected they are no longer supplying 
subtitles (that is, an episode title, not subtitles during the show!) so one 
can't even leave description out of the matching process.

Mike Perkins