Re: add log messages when replication slots become active and inactive (was Re: Is it worth adding ReplicationSlot active_pid to ReplicationSlotPersistentData?)

On 2023-Nov-16, Peter Smith wrote:

> I searched HEAD code and did not find any "translator:" comments for
> just ordinary slot name substitutions like these; AFAICT that comment
> is not necessary anymore.

True. Lose that.

The rationale I have is to consider whether a translator looking at the
original message message in isolation is going to understand what the %s
means. If it's possible to tell what it is without having to go read
the source code that leads to the message, then you don't need a
"translator:" comment. Otherwise you do.

You also need to assume the translator is not stupid, but that seems an
OK assumption.

> SUGGESTION (#1a and #1b)
>
> ereport(log_replication_commands ? LOG : DEBUG1,
> errmsg(SlotIsLogical(s)
> ? "acquired logical replication slot \"%s\""
> : "acquired physical replication slot \"%s\"",
> NameStr(s->data.name)));

The bad thing about this is that gettext() is not going to pick up these
strings into the translation catalog. You could fix that by adding
gettext_noop() calls around them:

ereport(log_replication_commands ? LOG : DEBUG1,
errmsg(SlotIsLogical(s)
? gettext_noop("acquired logical replication slot \"%s\"")
: gettext_noop("acquired physical replication slot \"%s\""),
NameStr(s->data.name)));

but at that point it's not clear that it's really better than putting
the ternary in the outer scope.

You can verify this by doing "make update-po" and then searching for the
messages in postgres.pot.

--
Álvaro Herrera PostgreSQL Developer — https://www.EnterpriseDB.com/
"Cada quien es cada cual y baja las escaleras como quiere" (JMSerrat)