[mythtv-users] Dup detection broken in some cases - how to fix?

[chris at cpr.homelinux.net]

I have noticed that a lot of the time when shows are current they 
have good subtitles and descriptions, but when they go into 
syndication some channels simply list the title without any other 
information.  With both 0.18 and 0.19, Myth will never consider 
those shows to have been "previously recorded" no matter what 
dup-matching method is in use.  If I see one in the listing and 
select "do not record" then Myth creates the "previously recorded" 
entry as expected and then *still* flags that episode for 
recording.  I could replace every recording rule with a power 
search that checks for subtitle>"" to prevent these syndicated 
reruns from getting into the list in the first place but that is a 
hassle (which takes away from the convenience of just seeing a show 
I like and adding it to the list) and wouldn't be required if Myth 
would recognize that ""=="" when comparing the listings to the old 
recordings.  When the shows go into syndication and there's no 
subtitle or description there's also usually no original broadcast 
date.

The other case where dupe detection breaks is when a show is listed 
on several channels and each one has a subtle difference in 
punctuation.  I end up recording multiples because "John gets a 
haircut." is different from "John gets a haircut".  Unlike the 
blank subtitles and descriptions, this time I don't want to exclude 
them from the list so a power rule won't help.  What's really 
needed is for Myth to either drop all non-alphabet characters prior 
to making the comparison or else use something like soundex 
comparisons (which would also catch some speeling errors).

Can anyone think of any short-term fixes for these problems (like 
options I've overlooked)?