> Bruce Momjian <pgman@candle.pha.pa.us> writes:
> > ! else if (strcasecmp(name, "all") == 0)
> > ! {
> > ! reset_DefaultXactIsoLevel();
> > ! reset_XactIsoLevel();
> > ! reset_random_seed();
> > ! /* reset_server_encoding(); */
> > ! reset_client_encoding();
> > ! reset_datestyle();
> > ! reset_timezone();
> > !
> > ! ResetAllOptions();
> > ! } else
>
> This looks fairly bogus, given the comment on ResetAllOptions:
>
> /*
> * Reset all options to their specified default values. Should only be
> * called at program startup.
> */
>
> Or is that OK now with the "makeDefault" addition to SetConfigOption?
> Peter, your thoughts please? (If it is OK, the patch needs to change
> this comment. Some additional commentary about what exactly the
> makeDefault switch does would be appropriate too, since it's not real
> clear exactly how "default" the value is becoming.)
>
> I also wonder whether it's a good idea to reset the client encoding
> here... it might be correct in a pure sense, but isn't it likely to
> break the client?
I assumed those were items not controlled by postgresql.conf like
isolation level. Also, do we have a way to clear transaction state? Is
BEGIN;COMMIT; the clean way to do it?
--
Bruce Momjian | http://candle.pha.pa.us
pgman@candle.pha.pa.us | (610) 853-3000
+ If your life is a hard drive, | 830 Blythe Avenue
+ Christ can be your backup. | Drexel Hill, Pennsylvania 19026