Re: BUG #17385: "RESET transaction_isolation" inside serializable transaction causes Assert at the transaction end

At Mon, 14 Mar 2022 17:12:18 +0900, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote in
> On Sat, Mar 12, 2022 at 4:53 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >
> > Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> writes:
> > > I've attached a draft patch for discussion.
> >
> > Hm, I think that trying to RESET a GUC_NO_RESET variable ought to
> > actively throw an error. Silently doing nothing will look like
> > a bug.
>
> Agreed. I've attached an updated patch.

FWIW,

+ /* Check if the parameter supports RESET */
+ if (record->flags & GUC_NO_RESET && value == NULL)
+ {
+ ereport(elevel,
+ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+ errmsg("parameter \"%s\" cannot be reset",
+ name)));

ERRCODE_FEATURE_NOT_SUPPORTED doesn't look proper for the case. Isn't
it ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE or something like?

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center