On 1/3/16 11:24 PM, Michael Paquier wrote:
>> Sorry, didn't realize you were on it.
> No worries. I know it's already late where you are.
And late where 9.5 is... ;)
> I would use != 1 instead here, even if the function is strict.
Yeah, I effectively pulled the pattern from DeconstructQualifiedName,
but != 1 is better.
> You forgot to update the regression test output. And actually on
Doh.
> second thoughts the tests you added do not actually bring much value
> because what is tested is just the behavior of
> stringToQualifiedNameList, and the other reg* are not tested that as
Seemed good to test what the original bug was, but sure.
> well. However I think that we had better test the failures of
> regnamespace and regrole when more than 1 element is parsed as this is
> a new particular code path. Attached is an updated patch which passes
> check-world in my environment.
Patch looks good to me. FWIW, RhodiumToad and macdice looked at my patch
as well and didn't see any problems you didn't mention.
--
Jim Nasby, Data Architect, Blue Treble Consulting, Austin TX
Experts in Analytics, Data Architecture and PostgreSQL
Data in Trouble? Get it in Treble! http://BlueTreble.com