Re: BUG #8633: Assigning to a variable named "current_time" gives wrong output

Tom Lane-2 wrote
> David Johnston <

> polobo@

> > writes:
>> bricklen wrote
>>> I don't think this is a bug, "current_time" is a reserved word. See
>>> http://www.postgresql.org/docs/current/static/sql-keywords-appendix.html
>
>> Agreed; though I am curious why this does not throw an error during the
>> declaration or assignment. A spurious parser error would be welcomed
>> compared to silently ignoring the requested action.
>
> Well, it's not "ignoring the action", it's just that the keyword meaning
> now takes precedence in expressions. We changed that in 9.0 I believe.
>
> We could possibly throw an error if you use a SQL reserved word in a
> declaration without double-quoting it. That wouldn't be a complete fix,
> because if you did that and then forgot to double-quote the name in
> expressions, it'd still do the wrong thing. But this might at least
> help you remember you need to do that.
>
> I think though that at one time it was considered a feature that we
> didn't insist on double-quoting plpgsql variable names unnecessarily.
> Don't know if changing this would be a net improvement for everyone.

If we had done this when all of the other pl/pgsql variable-name changes
were put in place that would have been nice but to do it stand-alone
definitely has a higher hurdle to clear.

At this point I'd rather leave this to a linter / static-analyzer rather
than affect existing code unconditionally.

David J.

--
View this message in context: http://postgresql.1045698.n5.nabble.com/BUG-8633-Assigning-to-a-variable-named-current-time-gives-wrong-output-tp5780601p5780626.html
Sent from the PostgreSQL - bugs mailing list archive at Nabble.com.