zeitgeist team mailing list archive

[Mikkel Kamstrup Erlandsen <mikkel.kamstrup@xxxxxxxxx>]

2009/10/8 Siegfried Gevatter <rainct@xxxxxxxxxx>:
> 2009/10/8 Seif Lotfy <seif@xxxxxxxxx>:
>> All I know there should not be any DB
>> changes except for removing the unused column "end" in the event table.
>
> And (at least) changing the content/source URIs.

You mean to use the Nepomuk ontologies?

> Mikkel, I like your new suggestion, although it will make the current
> huge FindEvents even more unreadable :P (we need to look if there's a
> better way to write that, like using procedures stored in SQLite or
> several queries, but it's not that easy as the available filters are
> exteremly powerful).

Yeah. The complexity in find_events() is indeed worrying... I have
thought about making a stand alone QueryCompiler class that tries to
do this as cleanly and self contained as possible. A big problem with
the current find_events() is that it is incredibly hard to guess
whether or not we use the indexes in the DB right (and for the record;
I am quite convinced we don't use the indexes very well now). Another
big problem is that it is virtually impossible to debug :-)

That said, I am not sure just how much more readable we can make the
find_events() code given the API we expose. This area clearly needs
some hard thinking and hard coding :-)

Stored procedures, as you say, might also make things a lot easier.
And as we discussed on IRC a while ago a few simple queries might
perform better than one very complex one.

-- 
Cheers,
Mikkel