Re: on placeholder entries in view rule action query's range table

On Thu, Jan 12, 2023 at 09:54:09AM -0500, Tom Lane wrote:
> Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
> > On 2023-01-12 Th 00:12, Justin Pryzby wrote:
> >> It's ugly and a terrible hack, and I don't know whether anyone would say
> >> it's good enough, but one could can probably avoid the diff like:
> >> sed -r '/CREATE/,/^$/{ s/\w+\.//g }'
>
> > That looks quite awful. I don't think you could persuade me to deploy it
> > (We don't use sed anyway). It might be marginally better if the pattern
> > were /CREATE.*VIEW/ and we ignored that first line, but it still seems
> > awful to me.
>
> Yeah, does not sound workable: it would risk ignoring actual problems.
>
> I was wondering whether we could store a per-version patch or Perl
> script that edits the old dump file to remove known discrepancies
> from HEAD. If well-maintained, that could eliminate the need for the
> arbitrary "fuzz factors" that are in TestUpgradeXversion.pm right now.
> I'd really want these files to be kept in the community source tree,
> though, so that we do not need a new BF client release to change them.
>
> This isn't the first time this has come up, but now we have a case
> where it's actually blocking development, so maybe it's time to
> make something happen. If you want I can work on a patch for the
> BF client.

What about also including a dump from an old version, too ?
Then the upgrade test can test actual upgrades.

A new dump file would need to be updated at every release; the old ones
could stick around, maybe forever.

--
Justin