Re: run pgindent on a regular basis / scripted manner

Peter Eisentraut  writes:
> On 24.04.23 16:14, Tom Lane wrote:
>> I certainly don't like its current behavior where adding/changing one
>> line can have side-effects on nearby lines.  But we have a proposal
>> to clean that up, and I'm cautiously optimistic that it'll be better
>> in future.  Did you have other specific concerns?

> I think the worst is how it handles multi-line data structures like

>          $newnode->command_ok(
>              [
>                  'psql', '-X',
>                  '-v',   'ON_ERROR_STOP=1',
>                  '-c',   $upcmds,
>                  '-d',   $oldnode->connstr($updb),
>              ],
>              "ran version adaptation commands for database $updb");

Yeah, I agree, there is no case where that doesn't suck.  I don't
mind it imposing specific placements of brackets and so on ---
that's very analogous to what pgindent will do.  But it likes to
re-flow comma-separated lists, and generally manages to make a
complete logical hash of them when it does, as in your other
example:

>          $node->command_fails_like(
>              [
>                  'pg_basebackup',   '-D',
>                  "$tempdir/backup", '--compress',
>                  $cft->[0]
>              ],
>              qr/$cfail/,
>              'client ' . $cft->[2]);

Can we fix it to preserve the programmer's choices of line breaks
in comma-separated lists?

			regards, tom lane