[mythtv-users] RFC: Allow mythfilldatabase to use a single XML source file for multiple video sources

[Nick]

 ''(Perhaps this only applies to XMLTV-based grabbers)''

 Currently I have 3 video sources on 5 cards (digital cable, digital
 terrestrial [DVB-T], and analogue terrestrial). All sources use the uk_rt
 XMLTV grabber for listings data. I currently have to run the grabber 3
 times, once for each source with  own config file, and run
 mythfilldatabase from cron on each of the 3 XML files produced. (The
 digital cable video source contains _nearly_ all channels found on the
 other 2 sources, so is almost a super-set of the 3 sources itself.)

 I would like the ability to run a single cron job for the grabber, and a
 single cron job for MFDB. The grabber xmltv configuration file would
 contain all of the XMLTV IDs for all of the unique channels on all
 sources. This would allow a 'super-set' of listings data to be created in
 a single file. MFDB would then be run on this single data file and would
 update listings data on all channels where there is a XMLTVID match. This
 could be achieved by passing a CSV-list of sourceids to MFDB at runtime.

 Herein lies the issue - I can easily get a single XML file containing all
 of the listings data I want, but MFDB will add new instances of channels
 on sources that do not contain channels matching all XMLTVIDs found in the
 source file. This intended behaviour therefore should be coupled with the
 abiity for MFDB to _not_ automatically create new channel instances on
 sources where not all XMLTVIDs are matched (something like --dont-add-
 channels).

 It seems inefficient (to me) to have run MFDB 3 times to import the exact
 same data onto 3 different sources having common channels, and also to
 have to hit the BBC's XML listings each time (even though I cache the
 listings data through Perl's httpcache).

 I know from the mythtv-users list that the auto creation of channels found
 in the source data but not in the DB also causes undue frustration, and I
 would relish seeing this as a separate run-time option for MFDB.

 Do any other users think this is a useful suggestion?

Cheers,
Nick

P.S. Apologies to the trac admins for posting this there.