Horton

Re: removing global variable ThisTimeLineID

Michael Paquier <michael@paquier.xyz>

View thread View on postgr.es
From: Michael Paquier <michael@paquier.xyz>
To: Robert Haas <robertmhaas@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-11-09T03:30:53Z
Lists: pgsql-hackers 
On Mon, Nov 08, 2021 at 12:49:52PM -0500, Robert Haas wrote:
> Even with this patch, the name ThisTimeLineID is still used for
> multiple purposes. It remains part of the CheckPoint struct, and also
> part of the xl_end_of_recovery struct. But in both of those cases, the
> name ThisTimeLineID actually makes sense, because what we're thinking
> about is whether there's a timeline change, so we have ThisTimeLineID
> and PrevTimeLineID and it seems fairly clear what that's supposed to
> mean.

I got to wonder if it would be better to bite the bullet here for
these two looking at your patch, but I am fine to keep things as they
are, as well.

> Thoughts?

I think that this patch is an improvement.

@@ -6686,8 +6682,8 @@ StartupXLOG(void)
-   TimeLineID  ThisTimeLineID,
-               PrevTimeLineID;
+   TimeLineID  replayTLI,
+               newTLI;
One problem with newTLI is that this conflicts with the declaration
around 7663 in xlog.c, where we check after a TLI switch when
replaying such a record.  Perhaps this could be named newInsertTLI,
for example.
--
Michael

Commits

  1. More cleanup of 'ThisTimeLineID'.

  2. Change ThisTimeLineID from a global variable to a local variable.

  3. Remove all use of ThisTimeLineID global variable outside of xlog.c

  4. Don't set ThisTimeLineID when there's no reason to do so.

  5. When fetching WAL for a basebackup, report errors with a sensible TLI.

  6. Remove useless code from CreateReplicationSlot.