2010/8/1 Robert Haas <robertmhaas@gmail.com>:
> On Sun, Jul 25, 2010 at 11:42 AM, Pavel Stehule <pavel.stehule@gmail.com> wrote:
>>> I'm setting this as ready for committer.
>>
>> Thank you very much
>
> I took a look at this tonight and am a bit mystified by the following bit:
>
> + /*
> + * PL doesn't calculate first row of function's body
> + * when first row is empty. So checks first row, and
> + * correct lineno when it is necessary.
> + */
>
> Is that true of any PL, or just some particular PL? Is the behavior
> described here a bug that we should fix, or is it, for some reason,
> considered correct? Is it documented in our documentation?
it is primary plpgsql issue.
>
> The implementation of first_row_is_empty() looks pretty kludgey, too.
> It seems to me that it will fail completely if the text of the
> function definition happens to contain $function$.
>
> CREATE OR REPLACE FUNCTION boom() RETURNS text LANGUAGE plpgsql AS $$
> BEGIN SELECT '$function$'; END $$;
>
I can enhance algorithm on client side - but it will not be a pretty
code - it better do it on server side - for example
pg_get_formated_functiondef ...
CREATE OR REPLACE FUNCTION pg_get_formated_function_def(in oid,
linenum bool:= false, OUT funcdef text, OUT first_line_isignored bool)
this can remove any ugly code
Regards
Pavel
> --
> Robert Haas
> EnterpriseDB: http://www.enterprisedb.com
> The Enterprise Postgres Company
>