Thread: `transaction_read_only` GUC undocumented

On Tue, Jan 22, 2019 at 01:26:40PM +0100, Pavlo Golub wrote:
> I found that `transaction_read_only`GUC is undocumented. The only
> place where it referenced is Hot Standby section
> (https://www.postgresql.org/docs/current/hot-standby.html)
>
> Can we add description for this GUC, please?

That makes sense, now the documentation includes only a short of
roughly what it is here for and how hot standby uses it.  Would you
like to write a patch for config.sgml?
--
Michael

On Wed, Jan 23, 2019 at 06:18:58AM +0200, Pavlo Golub wrote:
> Yeah, why not. Could you please give me a hint where can I find
> details about it? Sources or commits?

(You may want to avoid HTML..)

transaction_read_only is listed in src/backend/utils/misc/guc.c, so we
could put a description close to default_transaction_read_only in
config.sgml, except that transaction_read_only sets the current
transaction's read-only status.
--
Michael

Michael Paquier <michael@paquier.xyz> writes:
> transaction_read_only is listed in src/backend/utils/misc/guc.c, so we
> could put a description close to default_transaction_read_only in
> config.sgml, except that transaction_read_only sets the current
> transaction's read-only status.

Do we really want to document that?  It's not the standard-approved
way of making a transaction read-only.

            regards, tom lane

On Wed, Jan 23, 2019 at 01:07:31AM -0500, Tom Lane wrote:
> Michael Paquier <michael@paquier.xyz> writes:
>> transaction_read_only is listed in src/backend/utils/misc/guc.c, so we
>> could put a description close to default_transaction_read_only in
>> config.sgml, except that transaction_read_only sets the current
>> transaction's read-only status.
>
> Do we really want to document that?  It's not the standard-approved
> way of making a transaction read-only.

i'd rather document them clearly rather than letting them around with
users guessing what they actually do (remember the recent thread about
replication parameter in connection strings).  By the way, I can see
that transaction_deferrable and transaction_isolation are not around
either...
--
Michael

Greetings, Michael.

You wrote 23.01.2019, 8:14:

> On Wed, Jan 23, 2019 at 01:07:31AM -0500, Tom Lane wrote:
>> Michael Paquier <michael@paquier.xyz> writes:
>>> transaction_read_only is listed in src/backend/utils/misc/guc.c, so we
>>> could put a description close to default_transaction_read_only in
>>> config.sgml, except that transaction_read_only sets the current
>>> transaction's read-only status.
>> 
>> Do we really want to document that?  It's not the standard-approved
>> way of making a transaction read-only.

> i'd rather document them clearly rather than letting them around with
> users guessing what they actually do (remember the recent thread about
> replication parameter in connection strings).

Exactly my case. Client sent support request stating that changing
this parameter doesn't work. Of course, he was doing this outside of
transaction. And of course he doesn't find manual reference and was
guessing about it's behavior.

>  By the way, I can see
> that transaction_deferrable and transaction_isolation are not around
> either...

> --
> Michael



-- 
Kind regards,
 Pavlo                          mailto:pavlo.golub@cybertec.at