From: Fujii Masao <masao.fujii@oss.nttdata.com>
To: Masahiko Sawada <sawada.mshk@gmail.com>
To: Shinya11.Kato@nttdata.com
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: [PATCH] Feature improvement for CLOSE, FETCH, MOVE tab completion
Date: Thu, 7 Jan 2021 10:59:28 +0900
Message-ID: <a562a2cb-e2dd-9c02-d849-a52db9cf6907@oss.nttdata.com> (raw)
In-Reply-To: <CAD21AoDR5BiPjTpWsc-EmG0+_Gt43Poc0G3wjBpw1kgHhiSvWA@mail.gmail.com>
References: <b0e4c5c53ef84c5395524f5056fc71f0@MP-MSGSS-MBX001.msg.nttdata.co.jp>
<CAD21AoBRw9FY6oNj8ubY7XEwJ9Bprfehu88_ePbCfH93JLxWsw@mail.gmail.com>
<7559e8ee8a0644c8aad79837092143c7@MP-MSGSS-MBX001.msg.nttdata.co.jp>
<0eab839e-5827-e517-4abf-00b3e581cd45@oss.nttdata.com>
<CAD21AoBiGkvFpizGAWt-8mQEja3_Afu0pjn09vR1Gk7d+zj_Cw@mail.gmail.com>
<697532b0-0a83-7046-1a58-b1ee0f66ac9a@oss.nttdata.com>
<7bf2e1700a7345f79c04fd513675467a@MP-MSGSS-MBX001.msg.nttdata.co.jp>
<CAD21AoDR5BiPjTpWsc-EmG0+_Gt43Poc0G3wjBpw1kgHhiSvWA@mail.gmail.com>
On 2021/01/07 10:01, Masahiko Sawada wrote:
> On Wed, Jan 6, 2021 at 3:37 PM <Shinya11.Kato@nttdata.com> wrote:
>>
>>> +#define Query_for_list_of_cursors \
>>> +" SELECT name FROM pg_cursors"\
>>>
>>> This query should be the following?
>>>
>>> " SELECT pg_catalog.quote_ident(name) "\
>>> " FROM pg_catalog.pg_cursors "\
>>> " WHERE substring(pg_catalog.quote_ident(name),1,%d)='%s'"
>>>
>>> +/* CLOSE */
>>> + else if (Matches("CLOSE"))
>>> + COMPLETE_WITH_QUERY(Query_for_list_of_cursors
>>> + " UNION ALL SELECT 'ALL'");
>>>
>>> "UNION ALL" should be "UNION"?
>>
>> Thank you for your advice, and I corrected them.
>>
>>>> + COMPLETE_WITH_QUERY(Query_for_list_of_cursors
>>>> + " UNION SELECT 'ABSOLUTE'"
>>>> + " UNION SELECT 'BACKWARD'"
>>>> + " UNION SELECT 'FORWARD'"
>>>> + " UNION SELECT 'RELATIVE'"
>>>> + " UNION SELECT 'ALL'"
>>>> + " UNION SELECT 'NEXT'"
>>>> + " UNION SELECT 'PRIOR'"
>>>> + " UNION SELECT 'FIRST'"
>>>> + " UNION SELECT 'LAST'"
>>>> + " UNION SELECT 'FROM'"
>>>> + " UNION SELECT 'IN'");
>>>>
>>>> This change makes psql unexpectedly output "FROM" and "IN" just after "FETCH".
>>>
>>> I think "FROM" and "IN" can be placed just after "FETCH". According to
>>> the documentation, the direction can be empty. For instance, we can do
>>> like:
>>
>> Thank you!
>>
>>> I've attached the patch improving the tab completion for DECLARE as an
>>> example. What do you think?
>>>
>>> BTW according to the documentation, the options of DECLARE statement
>>> (BINARY, INSENSITIVE, SCROLL, and NO SCROLL) are order-sensitive.
>>>
>>> DECLARE name [ BINARY ] [ INSENSITIVE ] [ [ NO ] SCROLL ]
>>> CURSOR [ { WITH | WITHOUT } HOLD ] FOR query
>>>
>>> But I realized that these options are actually order-insensitive. For
>>> instance, we can declare a cursor like:
>>>
>>> =# declare abc scroll binary cursor for select * from pg_class;
>>> DECLARE CURSOR
>>>
>>> The both parser code and documentation has been unchanged from 2003.
>>> Is it a documentation bug?
>>
>> Thank you for your patch, and it is good.
>> I cannot find the description "(BINARY, INSENSITIVE, SCROLL, and NO SCROLL) are order-sensitive."
>> I saw "The key words BINARY, INSENSITIVE, and SCROLL can appear in any order." , according to the documentation.
>
> Thanks, you're right. I missed that sentence. I still don't think the
> syntax of DECLARE statement in the documentation doesn't match its
> implementation but I agree that it's order-insensitive.
>
>> I made a new patch, but the amount of codes was large due to order-insensitive.
>
> Thank you for updating the patch!
>
> Yeah, I'm also afraid a bit that conditions will exponentially
> increase when a new option is added to DECLARE statement in the
> future. Looking at the parser code for DECLARE statement, we can put
> the same options multiple times (that's also why I don't think it
> matches). For instance,
>
> postgres(1:44758)=# begin;
> BEGIN
> postgres(1:44758)=# declare test binary binary binary cursor for
> select * from pg_class;
> DECLARE CURSOR
>
> So how about simplify the above code as follows?
>
> @@ -3005,8 +3014,23 @@ psql_completion(const char *text, int start, int end)
> else if (Matches("DECLARE", MatchAny))
> COMPLETE_WITH("BINARY", "INSENSITIVE", "SCROLL", "NO SCROLL",
> "CURSOR");
> + /*
> + * Complete DECLARE <name> with one of BINARY, INSENSITIVE, SCROLL,
> + * NO SCROLL, and CURSOR. The tail doesn't match any keywords for
> + * DECLARE, assume we want options.
> + */
> + else if (HeadMatches("DECLARE", MatchAny, "*") &&
> + TailMatches(MatchAnyExcept("CURSOR|WITH|WITHOUT|HOLD|FOR")))
> + COMPLETE_WITH("BINARY", "INSENSITIVE", "SCROLL", "NO SCROLL",
> + "CURSOR");
This change seems to cause "DECLARE ... CURSOR FOR SELECT <tab>" to
unexpectedly output BINARY, INSENSITIVE, etc.
Regards,
--
Fujii Masao
Advanced Computing Technology Center
Research and Development Headquarters
NTT DATA CORPORATION
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-hackers@postgresql.org
Cc: masao.fujii@oss.nttdata.com, sawada.mshk@gmail.com, Shinya11.Kato@nttdata.com
Subject: Re: [PATCH] Feature improvement for CLOSE, FETCH, MOVE tab completion
In-Reply-To: <a562a2cb-e2dd-9c02-d849-a52db9cf6907@oss.nttdata.com>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox