Andreas Karlsson <andreas@proxel.se> writes:
> On 01/04/2016 01:09 AM, David Fetter wrote:
>> On Wed, Dec 30, 2015 at 01:21:06PM +0100, Andreas Karlsson wrote:
>>> Another completion which is currently missing but I am not sure if we should
>>> add or not is completing "ALTER|CREATE|DROP USER" with "MAPPING FOR", but
>>> since it might interfere with completing to username for "ALTER|DROP USER" I
>>> am not sure we want it. What do you think?
>> Is there a way to require some reasonable chunk--say, one that's
>> disambiguated from the name any known ROLE with LOGIN--of MAPPING
>> before completing with MAPPING FOR? I confess to not knowing how the
>> new system works in enough detail to know that off the top of my head.
> No, and while it would not be too hard to build it would not be worth
> doing just for this use case.
The way we've solved other similar cases would translate like this:
instead of the "query for user names" just returning user names, add
on "UNION 'MAPPING FOR'". So if you do
# alter user <TAB>
where you're now offered
alice joe postgres
you'd instead get
alice joe postgres MAPPING FOR
Dunno if it's worth the trouble though.
regards, tom lane