Draft release notes for next week's releases are up
On Fri, Aug 2, 2024 at 4:26 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
See
As usual, please send corrections/comments by Sunday.
For the "Prevent infinite loop in VACUUM" item, I noticed two things
1) In REL_17_STABLE, the behavior is an ERROR -- not an infinite loop.
Perhaps it is not worth mentioning the differing behavior because 17
hasn't GA'd yet, but I just wanted to bring this up since I wasn't
sure of the convention.
2) The note says <command>VACUUM</command>. This issue is, of course,
not limited to explicitly issued vacuum commands (i.e.
autovacuum-initiated vacuums can also encounter it). Again, I don't
know if this matters or what the convention is here.
- Melanie
Melanie Plageman <melanieplageman@gmail.com> writes:
For the "Prevent infinite loop in VACUUM" item, I noticed two things
Thanks for looking!
1) In REL_17_STABLE, the behavior is an ERROR -- not an infinite loop.
Perhaps it is not worth mentioning the differing behavior because 17
hasn't GA'd yet, but I just wanted to bring this up since I wasn't
sure of the convention.
Right, at this point v17 is not a target of these notes, so I just
documented what happens in <= 16. I did include the commit hashes
for v17/master in the comments, but that's mostly to reassure anyone
who looks that the issue was addressed in those branches.
2) The note says <command>VACUUM</command>. This issue is, of course,
not limited to explicitly issued vacuum commands (i.e.
autovacuum-initiated vacuums can also encounter it). Again, I don't
know if this matters or what the convention is here.
I think we usually just say VACUUM. If it affected autovacuum
differently, that might be worth documenting; but IIUC it's the
same.
regards, tom lane