On Tue, Aug 5, 2014 at 3:41 AM, Noah Misch <n...@leadboat.com> wrote: > This remains open for 9.4. Your proposal to revert the feature in 9.4 and fix > it in 9.5 sounds reasonable.
Ok, I've gone ahead and done this. I'm sorry for the delays and confusion. > On Thu, Jul 10, 2014 at 04:15:35PM +0100, Greg Stark wrote: >> On Mon, Jul 7, 2014 at 8:35 AM, Sergey Muraviov >> <sergey.k.murav...@gmail.com> wrote: >> > So what's wrong with the patch? >> > And what should I change in it for 9.5? >> >> Possibly nothing. The concern was tha it's modifying the output in >> cases where the output is not \expanded and/or not wrapped. Now I've >> mostly convinced myself that those cases should change to be >> consistent with the wrapped output but there was at least one tool >> depending on that format (check_postgres) so perhaps it's not >> worthwhile and having it be inconsistent would be safer. > > I did try psql-wrapped-expanded-fix-v5.patch with the tests Peter and I posted > upthread, and those tests now behave as they do in released versions. What > cases did you find that still change vs. 9.3? I was trying to build a spreadsheet of every combination of these options. It turns out that 4-dimensional spreadsheets are kind of awkward. I think the fundamental dilemma was the same that was discussed upthread. If wrapped expanded mode should have an extra space padding column for the wrap indicators then all expanded modes should have that column to be consistent since wrapping shouldn't change the amount of padding. But that means unrelated queries changes format in ways people weren't expecting. They were depending on the basically inconsistent formatting where expanded didn't have the same padding that non-expanded formats had which was only the case because nobody had anticiapted expanded format needing space for wrapping indicators. -- greg -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
