Hi folks, I am replying to Erik's mail on the life of a plugin he sent today. Since the description is quite exhaustive, I do not want to quote it. But I have some propositions:
* Firstly, the way plugins are found and loaded looks fine to me. It should be
possible to specify which directories will be scanned. Maybe the version
number which is used to check compability could be called something like 'ABI
version' because it is independent from the API version. As long as the
interfaces match, the plugin is usable.
* I agree that the plugin factories may remain in memory, since they would not
consume much space. Yet I would be a nice-to-have I unused modules would be
unloaded in a more mature version of licq2. Since licq is a tool which runs
everytime, it should not be too memory-consumant.
A different approach to get a list of all plugins and save memory, as
well, could be to parse only for the plugin names or only load some meta
information such as name and dependencies and then unload plugins which are
not used. A plugin can be found anytime the same way it was found on licq
startup, so there is no need to keep unused factories in memoy.
* For licq startup it would be nice to be able to specify plugins, which have
to be loaded, those which do not need to be loaded successfully and to get
some feedback which can be evaluated by some startup script, i.e. a return
value. This enables scripting actions if loading licq fails.
So far, my 2 cents. Keep the good work going!
Stefan
--
____
/ \
/ ^ ^ \ Stefan Haun
I \/ I [EMAIL PROTECTED]
II II http://www.tuxathome.de
II II
I /\/\ I
pgpMqtcZ6ywpM.pgp
Description: PGP signature
