On 12/04/10 17:52, Ben Martin wrote:
> A few things I do with the SQLite engine are analysing apache log files,
> or other separated files (like running awk with SQL as the query
> language), and interacting with web services as a table, eg, selecting
> data from a google spreadsheet and a local sqlite database in a single
> query. As a fair amount of the use cases I have are simply introducing
> and joining data from other sources into queries I think perhaps
> starting out with just a record set returning function would be a good
> thing.
>
> This nicely avoids the issues that the filesystem can't really do MVCC
> or transactions properly, but gives access to some more diverse data in
> a query for folks who want to perform such queries.
>
If that's all you're after, you may well be able to use the existing
COPY protocol to populate a temp table by outputting suitable csv from
your data source.
--
Craig Ringer
