[PATCH] Change in mythfilldatabase to make life easier for xmltv users,

[anonymous]

Hi,

I always find that configuring TV Card in MyhTV is difficult for non US citizen. What I'm usually do is:

1/ Do a first frequency scan to find the channel. 2/ Go to mythweb, and fill the xmltvid column by hand for each entry in Channel table. 3/ Launch mythfilldatabase to fill missing informations (icons, programs, etc...).

Perhaps miss something somewhere, but what hurts is Step 2 because you don't have a link between channel frequency and xmltv informations. The only link is the channel name that could be the same, or near the same.

What I propose is to modify void ChannelData::handleChannels. The goal is when we are handling a channel coming from the xmltv parser, to verify if we don't already have an entry in the Channel table without xmltvid but with the same name. If so, we can do an update of this entry, instead of doing an insert.

Laurent.

[laurent@…]

Patch to enhance xmltv channel import

[laurent@…]

Patch review to comply to coding standard,

[stuartm]

Don't have time to test this one.

[mdean]

Seems to be a dup of #9417, which has a different patch

[stuartm]

The problem I have with fuzzy matching is the potential for false positives which would be much more serious than false negatives here. The user would probably not notice the error until they started watching recordings and found the wrong channel was recorded. At least when requiring manual intervention this is unlikely to occur.

The threshold of 0.9 is high but I'd still like to see some stats on the number of false positives in testing given the size of the sample (DVB-S in Europe would be a decent and varied sample set). This has the potential to be more useful than the patch attached to #9417 but only if it achieves an extremely low or zero error rate.

Our preferred fix for this has always been for work to be done on 'lineups' at the xmltv end but this appears to have stalled. If xmltv was responsible for matching up channel names, frequencies or service ids to the xmltvid then the results would be much more reliable. Therefore this patch would only be an interim solution at best.

[stuartm]

Since I've just noticed that both tickets were opened by the same submitter with different patches I'm closing this ticket and any further submissions/replies should go to #9417. The fact that the author submitted a simpler patch without fuzzy matching 18 months after the first suggests that this version wasn't 100% satisfactory.