On tis, 2010-09-28 at 12:18 -0400, Tom Lane wrote:
> I'm inclined to think that that's not a fatal objection; it's not like
> we haven't felt free to change psql's output format before. As long as
> we don't back-patch this change, it should be no worse than other things
> we've done to third-party code without a backwards glance.
In the past, pg_regress used diff -b or -w, so making whitespace changes
in psql was not a problem.
> It would be good to get rid of this whitespace because (I believe) it is
> one of very few reasons for needing to have any trailing whitespace in
> git-controlled files. If we could get to a point where trailing
> whitespace in patches could be rejected automatically, it'd eliminate
> one small pet peeve.
You won't be able to programmatically forbid all trailing whitespace (at
least without additional arrangements) because of:
psql -c 'select 1 as a, null as b' | cat -Aa | b$
---+---$1 | $<===
Plus, there might be tests that check trailing space behavior or some
such, but I haven't looked for those.