Just wanted to share with y'all what I wanna do with plperl:
1) I want to implement database access from plperl script by providing a
perl module DBD::PgSPI which instead of using libpq interface to talk to
database would use SPI. Thus, certain client-side scripts could become
stored procedures with no change of code.
2) When that's done, it'll be possible to have an 'application server'
running as an frontend to postgresql. An external program would take SOAP
(or CORBA or something) calls and translate them to calls of plperl stored
procedures. Since perl procedures are able to return complex data
structures, and methods to marshal them are easily available, such
procedure can encapsulate business logic in one place. Also, an neat
possibility is client providing perl code to process things to the server
(where SQL just won't do).
3) Possibly, later, the 'external program' described above could be merged
into postgresql proper, to achieve additional speedup.
Let me know if this all makes sense.
-alex