[mythtv-users] subtitle matching rule not working

[Michael T. Dean]

On 08/15/2009 04:12 AM, Bill Williamson wrote:
> On Sat, Aug 15, 2009 at 5:19 PM, Michael T. Dean wrote:
>   
>> On 08/14/2009 08:10 PM, Bill Williamson wrote:
>>     
>>> We have a new channel in AU called "Go!" which is showing reruns.  I
>>> have a rule set up for Seinfeld as "any time any channel" and "match
>>> duplicates using subtitle."
>>>
>>> The show shows up in the list as:
>>> Seinfeld: "The Watch"
>>> and so on, so it has a subtitle.  Despite this it will NEVER match
>>> duplicates.
>>>
>>> How can I debug this?  I've also tried "subtitle then description"
>>> which made no changes.
>>>       
>> Have you actually recorded that episode with that subtitle, yet?  If not,
>> it's not yet a duplicate.  If so, I'd guess your old recorded episode and
>> the new program listings episode both have programid's (that differ), so
>> it's not seen as a duplicate.
> They don't match as dupes on the upcoming recording screen, and myth
> records all of them every time. (so I'll have 2-3 of each episode
> already recorded)
>
> Why should program ID matter when I tell it to match via subtitle?  I
> know that shepherd is very dicky with its programids...

Because programid, if it exists, is considered the ultimate arbiter of 
uniqueness, so it always overrides the chosen duplicate-check method 
when both the old recorded episode and the new program listings episode 
have a programid.  The programid is meant to be a unique key for each 
episode of a show.  If it's anything else, your grabber needs to be 
fixed to not provide a programid.

If, however, you use multiple sources of listings data (i.e. XMLTV on 
some channels and EIT on others), you're likely to have different 
programid's for the same episodes.  The work Stuart A. is doing on EIT 
custom authorities may actually help out in this case (in 0.22 and 
beyond), but I don't know for sure.  Until then, if you have 
"incompatible" data sources, you can always just post-process your XMLTV 
data (even with a simple sed script) to remove programid's before 
running mythfilldatabase on the data.

Mike