Re: [patch] \g with multiple result sets and \watch with copy queries

Corey Huinker wrote:

> I think that because it is more complicated than usual psql, we may want to
> comment on the intention of the tests and some of the less-than-common psql
> elements (\set concatenation, resetting \o, etc). If you see value in that
> I can amend the patch.

If the intentions of some tests appear to be unclear, then yes, sure.
I don't feel the need to explain the "how" though. The other
comments in these files say why we're testing such or such case, but
don't go beyond that.

> Are there any options on COPY (header, formats) that we think we should
> test as well?

There are COPY tests already in src/test/regress/sql/copy*.sql, which
hopefully cover the many combination of options.

For \g and \o the intention behind the tests is to check that the
query output goes where it should in all cases. The options that can't
affect where the results go are not really in scope.

FTR I started a followup thread on this at [1], to be associated to a
new CF entry [2]

[1]
https://www.postgresql.org/message-id/flat/25c2bb5b-9012-40f8-8088-774cb764046d%40manitou-mail.org

[2] https://commitfest.postgresql.org/40/4000/

Best regards,
--
Daniel Vérité
https://postgresql.verite.pro/
Twitter: @DanielVerite