On Thu, Jan 16, 2020 at 01:08:16PM -0500, Tom Lane wrote:
Tomas Vondra <tomas.von...@2ndquadrant.com> writes:
This is one of the patches already marked as RFC (since September by
Alvaro). Anyone interested in actually pushing it, so that it does not
fall through to yet another commitfest?
TBH, I think we'd be better off to reject it. This makes a nontrivial
change in a very long-standing psql behavior, with AFAICS no way to
get back the old semantics. (The thread title is completely misleading
about that; there's no "option" in the patch as it stands.) Sure,
in a green field this behavior would likely be more sensible ... but
that has to be weighed against the fact that it's behaved the way it
does for a long time, and any existing scripts that are affected by
that behavior have presumably deliberately chosen to use it.
I can't imagine that changing this will make very many people happier.
It seems much more likely that people who are affected will be unhappy.
The compatibility issue could be resolved by putting in the option
that I suppose was there at the beginning. But then we'd have to
have a debate about which behavior would be default, and there would
still be the question of who would find this to be an improvement.
If you're chaining together commands with \; then it's likely that
you are happy with the way it behaves today. Certainly there's been
no drumbeat of bug reports about it.
I don't know, really, I only pinged this as a CFM who sees a patch
marked as RFC for months ...
The current behavior certainly seems strange/wrong to me - if I send
multiple queries to psql, I'd certainly expect results for all of them,
not just the last one. So the current behavior seems pretty surprising.
I'm unable to make any judgments about risks/benefits of this change. I
can't imagine anyone intentionally relying on the current behavior, so
I'd say the patch is unlikely to break anything (which is not already
broken). But I don't have any data to support this ...
Essentially, I'm just advocating to make a decision - we should either
commit or reject the patch, not just move it to the next commitfest over
and over.
regards
--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
