Re: My honours project - databases using dynamically attached entity-properties
От | tomas@tuxteam.de |
---|---|
Тема | Re: My honours project - databases using dynamically attached entity-properties |
Дата | |
Msg-id | 20070316054238.GA28203@www.trapp.net обсуждение исходный текст |
Ответ на | Re: My honours project - databases using dynamically attached entity-properties (Andrew Dunstan <andrew@dunslane.net>) |
Ответы |
Re: My honours project - databases using dynamically attached
entity-properties
|
Список | pgsql-hackers |
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thu, Mar 15, 2007 at 12:49:10PM -0400, Andrew Dunstan wrote: > Gregory Stark wrote: > >"Ron Mayer" <rm_pg@cheapcomplexdevices.com> writes: > > > > > >>I have a system with many essentially user-defined fields, and was > >>thinking of creating something similar to an Array type and writing > >>some GIST indexes for it. [...] > >I think that's what hstore does. Honestly I'm not entirely certain though. > Does hstore nest? My impression is that it doesn't. Which might well not > matter, of course. If what you mean is to have "mappings of mappings" then no. Hstore implements a data type for a (finite) mapping (a set of key -> value pairs, think "hash" for perl folks), with operations like "H1 contains H2" (in the sense that all key-value pairs in H2 are also in H1) supported by an index. Keys and values are strings. But I don't see why it couldn't be extended to more structured data types (one of the good things of hstore at the moment is its surprising simplicity, because it provides a good example of what GiST is good for, therefore I'd understand if the authors don't wanted to add that much hair without need). In my view it's one of the cutest applications of GiST. Regards - -- tomás -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (GNU/Linux) iD8DBQFF+i5OBcgs9XrR2kYRAprYAJ46ecAnSNWnh+rizGdhWLutU7BphACdHweH k3rdsxmUzuxthwldXvJ3UKw= =+sT5 -----END PGP SIGNATURE-----
В списке pgsql-hackers по дате отправления: