David Fetter <david@fetter.org> writes:
> On Tue, Mar 02, 2010 at 11:02:54PM -0500, Jaime Casanova wrote:
>> On Tue, Mar 2, 2010 at 8:53 PM, Tom Lane <tgl@postgresql.org> wrote:
>>> Instead of trying (and failing) to allow <<label>> at the end of a DECLARE
>>> section, throw an error message saying explicitly that the label must go
>>> before DECLARE.
>>
>> with the docs saying another thing, this couldn't be considered as a
>> bug fix and therefore backpatchable
> +1 for back-patching.
No, I intentionally didn't back-patch that. The only benefit of the
change is throwing a useful error message for questionable syntax.
As against that we have the possibility of breaking code that works
now, if someone's got an incorrectly-placed label that they're not
actually referencing. I grant that that's not a very likely case;
but the error that the message would help people fix isn't very
likely either, as demonstrated by the fact that this never came up
before. We shouldn't be making changes in the behavior of back
branches for such things.
regards, tom lane