"James.FitzGibbon" wrote:
>
> Has anyone out there tried DBD::Sprite?
> ...
> The commit/rollback jumps out at me as being cool, since (AFAIK) DBD::CSV
> doesn't do that (AutoCommit is always 1 I think).
Currently that's true for both DBD::CSV and DBD::AnyData. However
DBD::AnyData supports in-memory tables so one can retrieve the data from
a file, massage the data en masse in memory, then either abandon it or
save it en masse to file which essentially is a commit/rollback
mechanism. Eventually I may add further support for explicit
transactions in both CSV and AnyData.
> interested in hearing anything people have to say.
I too would be interested to hear comparisons especially in the area of
supported SQL (comparing Sprite to the joins and other SQL supported via
the new SQL::Statement in DBD::AnyData and DBD::CSV) and also in the
area of DBI and SQL standards. My impression is that Sprite allows one
to do a bunch of cool things with regexes etc. but that some of these
make Sprite scripts incompatible with other DBI drivers, for example
this kind of thing would be quite convenient for a coder but would seem
to disallow transition from a script with the Sprite driver and one with
a different driver:
6) Quotes are not used around data.
7) It is not necessary to call the "$dbh->quote()" method
all the time in your sql.
Of course DBD::AnyData is guilty of this kind of thing also in some
areas, but AFAIK only as regards $dbh->func() calls, not other DBI calls
or SQL.
--
Jeff
