[mythtv-users] Mooting architecture for a DataDirect replacement
Jay R. Ashworth
jra at baylink.com
Fri Jun 22 13:37:33 UTC 2007
On Fri, Jun 22, 2007 at 02:25:39PM +1000, Peter Schachte wrote:
> > Sure. But NNTP and "believe the last signed entry you got for the
> > timeslot/channel" is even lower load. On both ends.
> If you read further in my posting I suggested something more like
> NNTP, where the server holds precomputed differences to be downloaded.
> I assume you intend each "newsgroup" in your system would be a single
> channel, and an "article" would be some update to the schedule for
> that channel (which may be an update to an existing program entry as
> well as an added entry).
You could leverage newsgroups for program sources, I suppose, yes. I
hadn't crawled quite that far.
> The main difference is that (by my assumption) you're keeping
> different channels separate, whereas I'm keeping everything for one
> area together. Whether the overhead of having to ask for each channel
> separately i your scheme is more or less than the waste in my scheme
> from some people downloading info about channels they don't want is
> not clear to me. But in both cases the big improvement comes from only
> downloading changes.
Sure. I was, originally, actually planning on just having one
newsfroup, and taking the efficiency hit of having machines have to
grab all the new headers to see what to get -- because I was planning
on having Supernews and the like carry it.
Jay R. Ashworth Baylink jra at baylink.com
Designer The Things I Think RFC 2100
Ashworth & Associates http://baylink.pitas.com '87 e24
St Petersburg FL USA http://photo.imageinc.us +1 727 647 1274
More information about the mythtv-users