robert.mcnamara at gmail.com
Wed Aug 11 18:34:28 UTC 2010
On Wed, Aug 11, 2010 at 11:30 AM, David Blain <MythTv at theblains.net> wrote:
> When I originally wrote the upnp stack, I had it use the videometadata
> table. Some time ago, someone else switched it to use the upnpmedia table.
> If I remember correctly this was done due to the fact that the
> videometadata table didn't contain data unless the MythVideo plugin was
> installed on a frontend (which not everyone has) and a scan was performed.
> Also, newly added video's weren't picked up unless by upnp clients unless
> MythVideo was opened and a new scan was performed.
On the other hand, getting access to the preference screen for setting the
directory for the upnp scanner to *scan* requires mythvideo to be installed,
too. So either way, you're installing mythvideo.
> I personally feel that the video, music, photos, etc... scanning/tracking
> should be part of the backend and not tied to the MythVideo plugin (I also
> think it should all be in one table with unique properties stored in related
> tables, but that's a different conversation) . I know there is a lot of
> work being done to integrate mythvideo, add Storage Group support and use
> the new FileSystemWatcher Qt class to detect new files instead of scanning,
> all of which will make the final implementation completely different.
Some more recent revelations about limitations to QFileSystemWatcher make it
less likely that we will use it for MythVideo scans, but the goal is
definitely to move all that scanning into the backend. Probably a breakout
of the mythvideo metadata functionalities into a libmythvideo or the like,
and some periodic scan if the filesystem watcher isn't going to work.
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mythtv-dev