On Jun 15, Giuseppe Corbelli <[EMAIL PROTECTED]> wrote:
> Perhaps the AkaName table is missing a: DatabaseIndex('person',
> name='idx_person)
Yep, and AkaTitle needs an index for movieID.
> I have some questions: [AkaName]: AFAIK most of the searches are
> done using the phonetic codes. Is it necessary to have an index on
> "name" and "imdbIndex" columns?
Right: it's completely unuseful for both AkaName and AkaTitle.
> [Name]: Suspicious about utility of indexes on 'name' and 'imdbIndex'
It's used to extract persons/movies references, searching for
exact matches over names and titles.
What I'm wondering is: can we create index only _only_ for
the 'name' ('title', in the Title table) column, leaving out
imdbIndex, productionYear and kind?
After all what really makes the difference is a person's name
or a movie's title; there are very few persons with identical
names or movies with identical title.
After that the list of people/movies with the given name/title
is retrieved using the index, I suppose that a simple comparison
over the other fields (not using indexes) would not take so long.
> As I've said before I'd suggest to avoid multi-column indexes,
Makes sense (and, by the way, I've still to completely undestand
the difference).
> I can fix this in the CVS once we agree on indexes definition.
You have green light! :-)
By the way, I've committed to the cvs a new version: ForeignKey
was completely removed (it's still untested).
--
Davide Alberani <[EMAIL PROTECTED]> [PGP KeyID: 0x465BFD47]
http://erlug.linux.it/~da/
_______________________________________________
Imdbpy-devel mailing list
Imdbpy-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/imdbpy-devel
