On Thu, Oct 30, 2025 at 6:17 PM Maxim Orlov <[email protected]> wrote: > On Thu, 30 Oct 2025 at 12:10, Heikki Linnakangas <[email protected]> wrote: >> >> >> Oh I see, the 'base' is not necessarily the base offset of the first >> multixact on the page, it's the base offset of the first multixid that >> is written to the page. And the (short) offsets can be negative. That's >> a frighteningly clever encoding scheme. One upshot of that is that WAL >> redo might get construct the page with a different 'base'. I guess that >> works, but it scares me. Could we come up with a more deterministic scheme? >> > Definitely! The most stable approach is the one we had before, which > used actual 64-bit offsets in the SLRU. To be honest, I'm completely > happy with it. After all, what's most important for me is to have 64-bit > xids in Postgres, and this patch is a step towards that goal.
Yes, but why can't we have an encoding scheme which would both be deterministic and provide compression? The attached is what I meant in [1]. It's based on v19 and provide deterministic conversion of each 8 of 64-bit offsets into a chunks containing 64-bit base and 7 of 24-bit increments. I didn't touch pg_upgrade code though. Links. 1. https://www.postgresql.org/message-id/CAPpHfdtPybyMYBj-x3-Z5%3D4bj_vhYk2R0nezfy%3DVjcz4QBMDgw%40mail.gmail.com ------ Regards, Alexander Korotkov Supabase
v21-0001-Use-64-bit-multixact-offsets.patch
Description: Binary data
v21-0002-Add-pg_upgarde-for-64-bit-multixact-offsets.patch
Description: Binary data
v21-0003-Add-test-for-64-bit-mxoff-in-pg_resetwal.patch
Description: Binary data
