Thread

Re: POC: make mxidoff 64 bits

Ashutosh Bapat <ashutosh.bapat.oss@gmail.com> — 2025-12-05T13:42:57Z

On Fri, Nov 28, 2025 at 6:35 PM Ashutosh Bapat
<ashutosh.bapat.oss@gmail.com> wrote:
>
>
> I will continue reviewing it further.
>

There is duplication of code/functionality between server and
pg_upgrade. With it we carry all the risks that come with
code/functionality duplication like the copies going out of sync.
There may be a valid reason to do that but it's not documented in the
comments. At-least both mutlixact_new.c and slru_io.c are not as well
commented as their server counterparts. I understand that the SLRU
code in the server deals with shared memory which is not needed in
pg_upgrade; pg_upgrade will not need more than one buffer in memory
and pg_upgrade code doesn't need to deal with lock and it can not deal
with locks. That means the code required by pg_upgrade is much simpler
than that on the server. But there's also non-trivial code which is
required in both the cases. WIll it be possible to extract parts of
slru.c which deal with IO into slru_io.c, make it part of the core and
then use it in pg_upgrade as well as slru.c? Or whether it's possible
to make SLRU use local memory? And throwing some FRONTEND magic to the
mix, we may be able to avoid duplication. Have we tried this or
something else to avoid duplication? Sorry, if this has been discussed
earlier. Please point me to the relevant discussion if so.

--
Best Wishes,
Ashutosh Bapat