Re: psql: Add role's membership options to the \du+ command

On 6/23/23 11:52 AM, David G. Johnston wrote:
> On Thu, Jun 22, 2023 at 5:08 PM Tom Lane <tgl@sss.pgh.pa.us 
> <mailto:tgl@sss.pgh.pa.us>> wrote:
> 
>     "Jonathan S. Katz" <jkatz@postgresql.org
>     <mailto:jkatz@postgresql.org>> writes:
>      > On 6/15/23 2:47 PM, David G. Johnston wrote:
>      >> Robert - can you please comment on what you are willing to
>     commit in
>      >> order to close out your open item here.  My take is that the
>     design for
>      >> this, the tabular form a couple of emails ago (copied here), is
>      >> ready-to-commit, just needing the actual (trivial) code changes
>     to be
>      >> made to accomplish it.
> 
>      > Can we resolve this before Beta 2?[1] The RMT originally advised
>     to try
>      > to resolve before Beta 1[2], and this seems to be lingering.
> 
>     At this point I kinda doubt that we can get this done before beta2
>     either, but I'll put in my two cents anyway:

[RMT Hat]

Well, the probability of completing this before the beta 2 freeze is 
effectively zero now. This is a bit disappointing as there was ample 
time since the first RMT nudge on the issue. But let's move forward and 
resolve it before Beta 3.

>     * I agree that the "tabular" format looks nicer and has fewer i18n
>     issues than the other proposals.
> 
> As you are on board with a separate command please clarify whether you 
> mean the tabular format but still with newlines, one row per grantee, or 
> the table with one row per grantor-grantee pair.
> 
> I still like using newlines here even in the separate meta-command.

(I'll save for the downthread comment).

> 
>     * Personally I could do without the "empty" business, but that seems
>     unnecessary in the tabular format; an empty column will serve fine.
> 
> 
> I disagree, but not strongly.
> 
> I kinda expected you to be on the side of "why are we discussing a 
> situation that should just be prohibited" though.

[Personal hat]

I'm still not a fan of "empty" but perhaps the formatting around the 
"separate command" will help drive a conclusion on this.

> 
>     * I also agree with Pavel's comment that we'd be better off taking
>     this out of \du altogether and inventing a separate \d command.
>     Maybe "\drg" for "display role grants"?
> 
> Just to be clear, the open item fix proposal is to remove the presently 
> broken (due to it showing duplicates without any context) "member of" 
> array in \du and make a simple table report output in \drg instead.
> 
> I'm good with \drg as a new meta-command.

[Personal hat]

+1 for a new command. The proposal above seems reasonable.

Thanks,

Jonathan