Philip Warner <pjw(at)rhyme(dot)com(dot)au> writes:
> Just wanted to check that this was a known problem:
> ERROR: fmgr_info: language 20322 has old-style handler
> this is from current CVS when trying to call a procedure written in plpgsql.
It's certainly not a known problem --- if it were, the plpgsql regress
test would be failing.
I have a nasty feeling that the reason for the failure is that you
restored a DB dump that declares the plpgsql call handler as language
'C' (correct for 7.0) not 'newC' (correct for current sources).
I don't see any easy way around this :-(.
Anyone have an idea how we can avoid importing a bad declaration from
an old dump?
BTW, I'm leaving in a few minutes and won't be home till Wed. evening,
so don't expect any thoughts from this quarter...
regards, tom lane
In response to
pgsql-hackers by date
|Next:||From: Larry Rosenman||Date: 2000-10-29 15:09:29|
|Subject: regression failure/UnixWare7.1.1/current sources/multibyte.|
|Previous:||From: Larry Rosenman||Date: 2000-10-29 14:33:43|
|Subject: Re: more multibyte/After TGL...|