Re: bugfix: --echo-hidden is not supported by \sf statements - Mailing list pgsql-hackers

From Pavel Stehule
Subject Re: bugfix: --echo-hidden is not supported by \sf statements
Date
Msg-id CAFj8pRDamxW=1E5myHzHpUisnwdW9Ed_Qo1d7pqEiYPMJVTkng@mail.gmail.com
Whole thread Raw
In response to Re: bugfix: --echo-hidden is not supported by \sf statements  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: bugfix: --echo-hidden is not supported by \sf statements
List pgsql-hackers
2013/2/23 Tom Lane <tgl@sss.pgh.pa.us>:
> Pavel Stehule <pavel.stehule@gmail.com> writes:
>> 2013/2/23 Stephen Frost <sfrost@snowman.net>:
>>> Why is that?  lookup_function_oid() only collects the oid to pass to
>>> get_create_function_cmd(), why not just issue one query to the backend?
>>> And use PSQLexec() to boot and get --echo-hidden, etc, for free?  And
>>> eliminate the one-off error handling for just this case?
>
>> yes, we can do it. There is only one issue
>
>> routines for parsing function signature in regproc and regprocedure
>> should be updated - and I would to get some agreement than I start to
>> do modify core.
>
> Uh ... you seem to be asking for a blank check to modify regprocedure,
> which is unlikely to be forthcoming.  Why do you think these things need
> to be changed, and to what?

If I understand well to request to remove "minimal_error_message" we
need to have fault tolerant regprocedure.

I am looking on this code now, and it is not easy as I though - there
are two possible errors: not found or found more - so returning
InvalidOid is not enough - and then some "new lookup" function is not
simple or is ugly - and I am not sure, so cost is less than benefit.
in this case.

Regards

Pavel



>
>                         regards, tom lane



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: New hooks for snapshot and transaction ID acquisition from external source
Next
From: Greg Stark
Date:
Subject: Re: Materialized views WIP patch