Bugfix and improvements in multixact.c
Maxim Orlov <orlovmg@gmail.com>
From: Maxim Orlov <orlovmg@gmail.com>
To: Postgres hackers <pgsql-hackers@lists.postgresql.org>
Date: 2024-06-14T13:56:44Z
Lists: pgsql-hackers
Attachments
- v1-0003-Switch-from-macro-functions-to-inline-in-multixac.patch (application/octet-stream) patch v1-0003
- v1-0002-Fix-using-of-MultiXactOffset-type-insted-of-Trans.patch (application/octet-stream) patch v1-0002
- v1-0001-Get-rid-of-misleading-xid-arg-name-in-multixact.patch (application/octet-stream) patch v1-0001
Hi! While working on a making multix xact offsets 64-bit [0] I've discovered a minor issue. The thing is that type 'xid' is used in all macro, but it doesn't correct. Appropriate MultiXactId or MultiXactOffset should be used, actually. And the second thing, as Heikki Linnakangas points out, args naming is also misleading. Since, these problems are not the point of thread [0], I decided to create this discussion. And here is the patch set addressing mentioned issues (0001 and 0002). Additionally, I made an optional patch 0003 to switch from macro to inline functions. For me, personally, use of macro functions is justified if we are dealing with different argument types, to make polymorphic call. Which is not the case here. So, we can have more control over types and still generate the same code in terms of speed. See https://godbolt.org/z/KM8voadhs Starting from O1 function is inlined, thus no overhead is noticeable. Anyway, it's up to the actual commiter to decide does it worth it or not. Again, this particular patch 0003 is completely optional. As always, any opinions and reviews are very welcome! [0] https://www.postgresql.org/message-id/flat/ff143b24-a093-40da-9833-d36b83726bdf%40iki.fi#61d5a0e1cf6ab94b0e8aae8559bc4cf7 -- Best regards, Maxim Orlov.
Commits
-
Convert confusing macros in multixact.c to static inline functions
- 0099b9408e8c 17.0 landed