Обсуждение: psql completion bugs with access methods

Hi all,

I have bumped into a couple of issues with psql completion for access
methods:
1) CREATE INDEX USING suggests both index and table AMs.
2) CREATE TABLE USING has no completion support, USING not being
included in the completion, and the follow-up table AMs are missing as
well.
3) CREATE ACCESS METHOD TYPE suggests only INDEX.

Attached is a patch to close the gap.  Thoughts?
--
Michael

Hi,

I'm not sure I understand starting 10 threads about approximately the
same topic. That seems purely confusing.

On 2019-06-01 15:10:07 -0400, Michael Paquier wrote:
> I have bumped into a couple of issues with psql completion for access
> methods:
> 1) CREATE INDEX USING suggests both index and table AMs.

Let's fix that.


> 2) CREATE TABLE USING has no completion support, USING not being
> included in the completion, and the follow-up table AMs are missing as
> well.
> 3) CREATE ACCESS METHOD TYPE suggests only INDEX.

I don't think these are bugs. I'm fine with adding those for 12, but I
don't think it's needed.

Greetings,

Andres Freund

On Sat, Jun 01, 2019 at 12:25:29PM -0700, Andres Freund wrote:
> I'm not sure I understand starting 10 threads about approximately the
> same topic. That seems purely confusing.

Well, each topic is separated IMO and has a separate patch, so I just
wanted to keep the discussion of each issue clear.

> On 2019-06-01 15:10:07 -0400, Michael Paquier wrote:
>> I have bumped into a couple of issues with psql completion for access
>> methods:
>> 1) CREATE INDEX USING suggests both index and table AMs.
>
> Let's fix that.
>
>> 2) CREATE TABLE USING has no completion support, USING not being
>> included in the completion, and the follow-up table AMs are missing as
>> well.
>> 3) CREATE ACCESS METHOD TYPE suggests only INDEX.
>
> I don't think these are bugs. I'm fine with adding those for 12, but I
> don't think it's needed.

I would just fix both.  Once you apply the filtering of access AMs for
indexes, the rest just makes sense to get done as well.  If you are
strongly opposed to that, I am fine not to fix it, but as we're on
it.
--
Michael

On 2019-06-01 15:41:29 -0400, Michael Paquier wrote:
> On Sat, Jun 01, 2019 at 12:25:29PM -0700, Andres Freund wrote:
> > I don't think these are bugs. I'm fine with adding those for 12, but I
> > don't think it's needed.
> 
> I would just fix both.  Once you apply the filtering of access AMs for
> indexes, the rest just makes sense to get done as well.  If you are
> strongly opposed to that, I am fine not to fix it, but as we're on
> it.

"I'm fine with adding those for 12", so no, I'm not strongly opposed.

Greetings,

Andres Freund

On Sat, Jun 01, 2019 at 12:44:05PM -0700, Andres Freund wrote:
> "I'm fine with adding those for 12", so no, I'm not strongly opposed.

OK, fixed this one for now.
--
Michael