From: Tom Lane <tgl@sss.pgh.pa.us>
To: =?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?= <ilmari@ilmari.org>
cc: "tanghy.fnst@fujitsu.com" <tanghy.fnst@fujitsu.com>,
        Peter Eisentraut <peter.eisentraut@enterprisedb.com>,
        Japin Li <japinli@hotmail.com>,
        Kyotaro Horiguchi <horikyota.ntt@gmail.com>,
        Laurenz Albe <laurenz.albe@cybertec.at>,
        "smithpb2250@gmail.com" <smithpb2250@gmail.com>,
        "david.zhang@highgo.ca" <david.zhang@highgo.ca>,
        "pgsql-hackers@lists.postgresql.org"
 <pgsql-hackers@lists.postgresql.org>
Subject: Re: Support tab completion for upper character inputs in psql
In-reply-to: <8735l41ynm.fsf@wibble.ilmari.org>
References: 
 <OS0PR01MB611303B647967B6949B759F1FB4E9@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <OS0PR01MB6113CA04E06D5BF221BC4FE2FB429@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <OS0PR01MB61136721363BD08576725F22FB089@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <80b093ae-88bd-8d8a-68b9-b535a3ab11be@enterprisedb.com>
 <OS0PR01MB61137C7B2B09C0431415B9DBFBD69@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <54bd7839-f463-ef03-f2d2-ccb8184b9b27@enterprisedb.com>
 <1068530.1641480996@sss.pgh.pa.us>
 <TYCPR01MB6128BD8EA1529F4270114C18FB4D9@TYCPR01MB6128.jpnprd01.prod.outlook.com>
 <MEYP282MB1669EAF2183E67EAF86F7F61B64D9@MEYP282MB1669.AUSP282.PROD.OUTLOOK.COM>
 <TYCPR01MB6128D4933189A74AAD326F98FB4D9@TYCPR01MB6128.jpnprd01.prod.outlook.com>
 <3547066.1642272686@sss.pgh.pa.us>
 <OS0PR01MB6113C130461FB8A9DBA0DAEBFB5A9@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <OS0PR01MB6113DB08D8CEA4E214DA37BBFB5F9@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <2304294.1643152287@sss.pgh.pa.us> <2645164.1643315022@sss.pgh.pa.u!
 s>
 <OS0PR01MB611360D58EA41FC6557D92C0FB229@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <2771362.1643385796@sss.pgh.pa.us> <2805502.1643405157@sss.pgh.pa.us>
 <2818574.1643408192@sss.pgh.pa.us>
 <OS0PR01MB61130E73CFDB2C0E8FD1A892FB249@OS0PR01MB6113.jpnprd01.prod.outlook.com>
 <324281.1643567679@sss.pgh.pa.us> <8735l41ynm.fsf@wibble.ilmari.org>
Comments: In-reply-to =?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?=
 <ilmari@ilmari.org>
	message dated "Mon, 31 Jan 2022 01:28:45 +0000"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <534640.1643594097.1@sss.pgh.pa.us>
Date: Sun, 30 Jan 2022 20:54:57 -0500
Message-ID: <534641.1643594097@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/534641.1643594097%40sss.pgh.pa.us>
Precedence: bulk

=?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?= <ilmari@ilmari.org> writes:
> First, as noted in the test, it doesn't preserve the case of the input
> for keywords appended to the query result.  This is easily fixed by
> using `pg_strdup_keyword_case()`, per the first attached patch.

I thought about that, and intentionally didn't do it, because it
would also affect the menus produced by tab completion.  Currently,
keywords are (usually) visually distinct from non-keywords in those
menus, thanks to being upper-case where the object names usually
aren't:

regression=# create table foo (c1 int, c2 int);            
CREATE TABLE
regression=# alter table foo rename c<TAB>
c1          c2          COLUMN      CONSTRAINT  

With this change, the keywords would be visually indistinguishable
from the object names, which I felt wouldn't be a net improvement.

We could do something hacky like matching case only when there's
no longer any matching object names, but that might be too magic.

> The second might be more of a matter of style or opinion, but I noticed
> a bunch of `if (foo) free(foo);`, which is redundant given that
> `free(NULL)` is a no-op.  To simplify the code further, I also made
> `escape_string(NULL)` be a no-op, returning `NULL`.

Yeah.  Our fairly longstanding convention is to avoid doing
free(NULL), dating back to when some platforms would crash on it.
I realize that's archaic now, but I'm not inclined to change
it in just some places.

			regards, tom lane