Обсуждение: Fixing tab-complete for dollar-names
Hi hackers,
In modern versions of Postgres the dollar sign is a totally legal character for identifiers (except for the first character), but tab-complete do not treat such identifiers well.
For example if one try to create an Oracle-style view like this:
create view v$activity as select * from pg_stat_activity;
, he will get a normally functioning view, but psql tab-complete will not help him. Type "v", "v$" or "v$act" and press <TAB> - nothing will be suggested.
Attached is a small patch fixing this problem.
In modern versions of Postgres the dollar sign is a totally legal character for identifiers (except for the first character), but tab-complete do not treat such identifiers well.
For example if one try to create an Oracle-style view like this:
create view v$activity as select * from pg_stat_activity;
, he will get a normally functioning view, but psql tab-complete will not help him. Type "v", "v$" or "v$act" and press <TAB> - nothing will be suggested.
Attached is a small patch fixing this problem.
Honestly I'm a little surprised that this was not done before. Maybe, there are some special considerations I am not aware of, and the patch will break something?
What would you say?
What would you say?
--
best regards,
Mikhail A. Gribkov
Mikhail A. Gribkov
Вложения
Hi hackers,
--
best regards,
Mikhail A. Gribkov
Mikhail A. Gribkov
On Sat, Jun 17, 2023 at 12:51 AM Mikhail Gribkov <youzhick@gmail.com> wrote:
Hi hackers,
In modern versions of Postgres the dollar sign is a totally legal character for identifiers (except for the first character), but tab-complete do not treat such identifiers well.
For example if one try to create an Oracle-style view like this:
create view v$activity as select * from pg_stat_activity;
, he will get a normally functioning view, but psql tab-complete will not help him. Type "v", "v$" or "v$act" and press <TAB> - nothing will be suggested.
Attached is a small patch fixing this problem.Honestly I'm a little surprised that this was not done before. Maybe, there are some special considerations I am not aware of, and the patch will break something?
What would you say?--best regards,
Mikhail A. Gribkov
On 6/26/23 22:10, Mikhail Gribkov wrote: > Hi hackers, > > As not much preliminary interest seem to be here, I'm sending the patch to > the upcoming commitfest I have added myself as reviewer. I already had taken a look at it, and it seemed okay, but I have not yet searched for corner cases. -- Vik Fearing
On 27/06/2023 02:47, Vik Fearing wrote: > On 6/26/23 22:10, Mikhail Gribkov wrote: >> Hi hackers, >> >> As not much preliminary interest seem to be here, I'm sending the patch to >> the upcoming commitfest > > I have added myself as reviewer. I already had taken a look at it, and > it seemed okay, but I have not yet searched for corner cases. LGTM, pushed. I concur it's surprising that no one's noticed or at least not bothered to fix this before. But I also couldn't find any cases where this would cause trouble. -- Heikki Linnakangas Neon (https://neon.tech)