Re: Reducing the WAL overhead of freezing in VACUUM by deduplicating per-tuple freeze plans

On Mon, Jan 9, 2023 at 1:43 PM Andres Freund <andres(at)anarazel(dot)de> wrote:
> ISTM that some of the page level freezing functions are misnamed. In heapam.c
> the heap_xlog* routines are for replay, afaict. However
> heap_xlog_freeze_plan() is used to WAL log the freeze
> plan. heap_xlog_freeze_page() is used to replay that WAL record. Probably your
> brain is too used to nbtree/ :).

Sometimes I wonder why other people stubbornly insist on not starting
every function name with an underscore. :-)

> I think s/heap_xlog_freeze/heap_log_freeze/ would mostly do the trick, except
> that heap_xlog_new_freeze_plan() doesn't quite fit in the scheme.

> The routines then also should be moved a bit up, because right now they're
> inbetween other routines doing WAL replay, adding to the confusion.

I believe that I used this scheme because of the fact that the new
functions were conceptually related to REDO routines, even though they
run during original execution. I'm quite happy to revise the code
based on your suggestions, though.

> The memcpy in heap_xlog_freeze_page() seems a tad odd. I assume that the
> alignment guarantees for xl_heap_freeze_plan are too weak?

They're not too weak. I'm not sure why the memcpy() was used. I see
your point; it makes you wonder if it must be necessary, which then
seems to call into question why it's okay to access the main array as
an array. I can change this detail, too.

I'll try to get back to it this week.

--
Peter Geoghegan