It's quiet...

[zollymonsta]

We're all silent in anticipation of the showdown at high noon... :-)

Peanut remotes at 50 paces?
Winner takes losers TiVo suggestions?

[zollymonsta]

*plies everyone with some steak and cheese pies*

[ehintz]

*plies everyone with some steak and cheese pies*

Me, I'd rather have a wee bit o' titty... (leers)

[thomson]

So... are people interested in having more than a few days worth of guide data? I am currently playing with a mechanism that duplicates information from the current week, so that I can have more than two weeks worth of data loaded. Using this mechanism the predicted data should be updated automatically as soon as new slices are released.

Currently this requires you to manually create/load slices... but if others like the idea, then we may want to incorporate it into the "official" slice.

[ehintz]

Sounds like a decent idea to me...

[STu]

Yes absolutely.
Additionally, when the real data did arrive would it be possible to have it retain the predictions where there were gaps in the real data?

I'd been looking at doing this with scripts to use last weeks guide data to fill any gaps in this this weeks data but doing it at the head end would obviously be preferable.

[thomson]

Additionally, when the real data did arrive would it be possible to have it retain the predictions where there were gaps in the real data?

Yeah, well... that is definitely the ultimate goal... but not as easy as it sounds. Needs some pretty heavy logic and parsing in order to replace the "gaps".
While this is easier if there are true gaps in the data... if the programme preceeding the gap has had its duration increased then this become much harder to diagnose. For example if a show that is usually only half and hour, is suddenly three hours, then it would be hard to figure out what the correct duration should be (as it could be a special episode)... if, however, the show was still only an 30 minutes long, but their was a two and a half hour gap with nothing showing, then this would be easier to match and fill in.

Still a pretty big job as the slice would need a lot of manipulation, rather than the simpler "search and replace" that I am currently doing.

[zollymonsta]

Me, I'd rather have a wee bit o' titty... (leers)

Oh of course :P

[STu]

While this is easier if there are true gaps in the data... if the programme preceeding the gap has had its duration increased then this become much harder to diagnose.

I take your point. I'd been working on some tcl to just fill the "gaps" from previously logged guide data and hadn't considered durations being extended. Does this actually happen? I'd be happy with just loosing the true gaps with all the "upcoming show" gaps getting filled with whatever show happened to be on at that time the previous week. shows at either end of the gap where overlaps potentially occur could just be dropped (I'm intending on putting in a truncated flag, easier when your on the client, to give the option).

Just need a second TiVo so I don't keep annoying the wife when I stuff the recordings index.

Anyways, take the simpler approach first and I'm sure we'll all appreciate whatever you come up with as always.

[thomson]

The easiest option for the gaps is to fix the bug in the slice generator that is creating them. I think the gap on TV2 on Sunday evenings has appeared more then twice now, but the TVNZ site seems fine and another source of FTA/XML information has the correct schedule. So I have a feeling the problem is likely to be in the listings->slice conversion process.

Now that the two weeks predictive information seems to be working for me, I will have a play at detecting the gaps and trying to fill in the information from another source... I was hoping to play with the predictive stuff to make the code more efficient. It should be noted that currently all of my filtering/parsing/predicting processing is being done on a 3GHz linux system rather than a 50MHz TiVo .