[mythtv] NEW MythPlugin functionality (PATCH)

[Edward Wildgoose]

Why not load up the nodes only when you initialise and then load the songs as the user opens up each node?  This assumes that the lag is proportional to quantity of retrieved data, rather than the time taken to draw the nodes say.

SELECT DISTINCT Artist, Album
FROM MusicTable

My MS SQL server scans over 200K rows per second on an 800Mhz machine (I haven't benchmarked for a while, that number might be on my older 200Mhz machine).  But either way doing a select distinct should be a fast operation in theory.

I haven't looked at the code, but assuming you have a flat file data structure then you could reduce "in-memory" requirements by normalising and replacing album/artist name with an ID field and having a seperate lookup table.  This shouldn't be slow in practice and might even speed things up (If it's already done this way then please ignore...)

If you get really desperate, then load the structure once, and then freeze/thaw it into a compressed structure when you aren't using it.  (ie via the plugin).  Initialisation then becomes the process of decompressing and extracting your data structure. (wacky, but hey...)

-----Original Message-----
From: Joseph A. Caputo [mailto:jcaputo1 at comcast.net]
Sent: 11 April 2003 16:47
To: Development of mythtv
Subject: RE: [mythtv] NEW MythPlugin functionality (PATCH)

It might be worth looking at these
apps to see how they do things, but I'm afraid in the final analysis that
there won't be that much we can do to reduce startup time short of keeping
the module resident.  If that's the case, we'll probably need to look into
reducing the constant memory requirements.

Unless everybody but me is happy with the status quo...