[mythtv-users] Zap2it Labs: Temporary scale-back to 3 days of listings

[Zap2It-Labs]

> > We update the database once a day, so the time of day you download 
> > within the 24 hour span is immaterial.  You won't get 
> anything new at 5:00 p.m.
> > (just before primetime) as opposed to 5:00 a.m.
> 
> Hmmm.  I can see how, no matter what time of day that update 
> actually *is*, it could interact poorly with a) when people 
> retrieve and b) when networks update.  Cause, no, I don't 
> think it actually *is* immaterial when we download.  If it's 
> *just before* you update, we lose an extra day's currency.

Yes, you're right.  I guess I'm used to some people downloading hundreds (or
even thousands) or times per day, totally pointlessly, from DataDirect,
thinking they're getting updated data (I guess).

One thing we may do, especially when Myth 0.19 comes out and becomes widely
adopted, is adjust the suggestedDownloadTimes returned by acknowledge() to
ensure that you'll always get a time after we've performed the data load.
For example, if we did the data load at 6:00 a.m., we wouldn't give suggest
a download time to Myth 0.19+ users of midnight to 6 a.m., since that would
be yesterday's stuff.

That's just an idea, not a policy.

 
> > Speaking somewhat off-the-cuff, if we do make a real change to the 
> > technical approach of DataDirect, it would probably be to 
> offer less 
> > customization and allow us to do more caching.  This would 
> cut down on 
> > CPU utilization (or allow us to "pre-calculate"), rather than doing 
> > expensive diff-type calculations for every user.
> 
> Hmmm.  Is there discussion going on in detail, anywhere, on 
> this topic?

Best place at this moment be this thread on Z2L:
http://bb.labs.zap2it.com/viewtopic.php?t=794


> And for what it's worth (just so my position is clear), you 
> guys *rock* for doing this at all (much less for free).  :-)  
> I'd just like to help all involved to make it work even 
> better on both sides.

Thanks. :)

Tom


> 
> Cheers,
> -- jra
> -- 
> Jay R. Ashworth