My wish list has one major item, removal of the tuple/record limit.
This limit renders the spatial functionality as totally inadequate for any serious use. I like the approach in some RDBMS's where there is a facility for "overflow" data, where the record size may have a basic limit, much as PostgreSQL has, but additional data for a record (tuple) may be stored in secondary "volumes". Configuration of these limits is able to be adjusted dynamically by the DBA to suit the record structure & optimise for useage.
Close to this in priority for my systems are referential integrity (esp foreign keys).
Some user control of index type would be nice also, not so much from a technical perspective, but a users: eg, a "time series" style index where each attr is unique (typically incrementing) & an index suitable for bucket hits, where an attr is indexed but not unique. A good quad-tree would be nice for polygons too....
Summat that would be nice is a "user supplied code" site on the WWW somwhere. (I haven't been able to find such a beast). Even some commercial software companies provide this type of servicefor users, full of disclaimers about the contents of course.
EG: I may well want a user defined function as an addition to SQL functionality to return the distance between two spatial entities (either cartesian or great circle). If there was a repository for such code I could just grab it, of it it wasn't in the repository I could write & upload it...
I believe that some of these are in the pipeline... Can't come too soon :-)
Join AllAdvantage.com and get paid to surf the Web!
Please use my ID (EIS-467) when asked if someone referred you. Thanks!
Get your own Free E-mail!
pgsql-general by date
|Next:||From: Kevin Lo||Date: 1999-12-26 09:42:10|
|Subject: Re: [GENERAL] DBI::Pg for Perl?|
|Previous:||From: john huttley||Date: 1999-12-26 08:00:06|
|Subject: Re: [GENERAL] Future of PostgreSQL|