Tom Lane <tgl@sss.pgh.pa.us> said:
> Seems like the only way to do that in the backend would be to find a way
> of slipping the function text past the lexer/parser entirely. While I
> can imagine ways of doing that, I think it'd be a *whole* lot cleaner
> to fix things on the client side.
>
> How do you feel about a psql hack that provides a "function definition"
> mode? More generally it could be a mode to enter random text and have
> it be converted to an SQL literal string. Perhaps
>
> psql=> create function foo (int) returns int as
> psql-> \beginliteral
> psql-LIT> begin
> psql-LIT> x := $1;
> psql-LIT> ...
> psql-LIT> end;
> psql-LIT> \endliteral
> psql-> language plpgsql;
>
> Essentially, \beginliteral and \endliteral each convert to a quote
> mark, and everywhere in between quotes and backslashes get doubled.
> We might want to specify that the leading and trailing newlines get
> dropped, too, though for function-definition applications that would
> not matter.
Tom --
Given that 98% of my function defining is done is psql, this would be fine for me and solve my frustrations. It
wouldn'thelp people that build functions in scripting languages or non-psql environments, however, but I don't know how
commonthis is.
What do others think?
Thanks!
--
Joel BURTON | joel@joelburton.com | joelburton.com | aim: wjoelburton
Knowledge Management & Technology Consultant
