Re: Query about time zone patterns in to_char

On Fri, Jul 9, 2021 at 10:44 AM Tomas Vondra
<tomas(dot)vondra(at)enterprisedb(dot)com> wrote:
> Yeah, does not seem to be worth it, as there seem to be no actual
> reports of issues in the field.
>
> FWIW there seem to be quite a bit of other to_char differences compared
> to Oracle (judging by docs and playing with sqlfiddle). But the patch
> seems fine / simple enough and non-problematic, so perhaps let's just
> get it committed?

This patch is still in the current CommitFest, so I decided to review
it. I see that DCH_keywords[] includes upper and lower-case entries
for everything except the three cases corrected by this patch, where
it includes upper-case entries but not the corresponding lower-case
entries. It seems to make sense to make these three cases consistent
with everything else.

It took me a while to understand how DCH_keywords[] and DCH_index[]
actually work, and I think it's a pretty confusing design, but what
the patch does seems to be consistent with that, so it appears correct
to me.

Therefore, I have committed it.

--
Robert Haas
EDB: http://www.enterprisedb.com