Alvaro Herrera wrote:
> With this patch applied, doing
> \h ALTER ROLE
> in psql looks quite odd: note how wide it has become. Maybe we should
> be doing this differently? (Hmm, why don't we accept ALL in the first SET
> line? Maybe that's just a mistake and the four lines should be all
> identical in the first half ...)
I have fixed the remaining issues, completed the doc changes, and
pushed. Given the lack of feedback I had to follow my gut on the best
way to change the docs. I added the regression test you submitted with
some additional changes, mainly to make sure they don't fail immediately
when other databases exist; maybe some more concurrency or platform
issues will show up there, but let's see what the buildfarm says.
Thanks Horiguchi-san for the patch and everyone for the reviews. (It's
probably worthwhile giving things an extra look.)
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services