Postgres Pro
Downloads
Home   >   mailing lists

Re: Schema variables - new implementation for Postgres 15 - Mailing list pgsql-hackers

From Pavel Stehule
Subject Re: Schema variables - new implementation for Postgres 15
Date
Msg-id CAFj8pRB9hqejftV4gdW2ovV7h_uM=e3rWZyq4TARi1FxefcTPw@mail.gmail.com
Whole thread Raw
In response to Re: Schema variables - new implementation for Postgres 15  (Dmitry Dolgov <9erthalion6@gmail.com>)
Responses Re: Schema variables - new implementation for Postgres 15  (Dmitry Dolgov <9erthalion6@gmail.com>)
List pgsql-hackers
Tree view


so 18. 11. 2023 v 15:54 odesílatel Dmitry Dolgov <9erthalion6@gmail.com> napsal:
> On Sat, Nov 18, 2023 at 02:19:09PM +0100, Pavel Stehule wrote:
> > Would it be a problem to make pg_session_variables inspect the catalog
> > or something similar if needed?
> >
>
> It can be very easy to build pg_session_variables based on iteration over
> the system catalog. But I am not sure if we want it. pg_session_variables()
> is designed to show the variables from session memory, and it is used for
> testing. Originally it was named pg_debug_session_variables. If we iterate
> over catalog, it means using locks, and it can have an impact on isolation
> tests.

I see, thanks for clarification. In the end one can check the catalog
directly of course, is there any other value in this function except for
debugging purposes?

I have no idea how it can be used for different purposes.  Theoretically it can be used to check if some variable was used (initialized) in a session already. But for this purpose it is not too practical, and if there will be some request for this functionality, then we can write a special function for this purpose. But I don't know any actual use cases for this.


As a side note, I'm intended to go one more time through the first few
patches introducing the basic functionality, and then mark it as ready
in CF. I can't break the patch in testing since quite long time, and for
most parts the changes make sense to me.

Thank you very much, for testing, comments, and all other work.

I marked pg_session_variables function as PARALLEL RESTRICTED, and did rebase

Regards

Pavel


Attachment

pgsql-hackers by date:

Previous
From: Joseph Koshakow
Date:
Subject: Re: Infinite Interval
Next
From: Andres Freund
Date:
Subject: Re: Use of backup_label not noted in log