(I know Daniel mentioned he'd get to these, but the ScanDirection one
was my fault and I needed to clear that off my mind. I did a few
others while on this topic.)
On Fri, 5 Jul 2024 at 05:00, Alexander Lakhin <exclusion@gmail.com> wrote:
> Please look at another bunch of inconsistencies/orphaned entities I found
> in the tree, with the possible substitutions:
> errmsg_buf -> errormsg_buf
> (coined by 6b18b3fe2)
Fixed.
> NoMovementScanDirectionScans -> NoMovementScanDirection
> (introduced with e9aaf0632, discussed in [1], but still seems inaccurate)
Oops. Fixed.
> XLogReadRecordInternal -> XLogReadRecord
> (from 3f1ce9734, align with a comment above: "Start and end point of last
> record returned by XLogReadRecord().")
Fixed.
> BYPASS_ALLOWCONN -> BGWORKER_BYPASS_ROLELOGINCHECK (see 492217301)
Fixed
> xs_ctup.t_self -> xs_heaptid (see c2fe139c2 and 304532421)
Fixed.
> pgStatShmLookupCache -> pgStatLocal.shmem (coined by 5891c7a8e)
Fixed.
> smgr_fsm_nblocks and smgr_vm_nblocks -> smgr_cached_nblocks
> (see the same comment updated by c5315f4f4)
Heikki fixed in 19de089cd.
> XID becomes older than GlobalXmin -> XID becomes visible to everyone
> (in accordance with dc7420c2c9 src/backend/access/gist/gistutil.c)
I'd need to spend more time to understand this.
> gen-rtab - remove (non-existing since db7d1a7b0)
Daniel fixed in cc59f9d0f.
> BARRIER_SHOULD_CHECK - remove (unused since a3ed4d1ef)
I wasn't sure if nothing else external could be using this. The macro
doesn't use any fields that are now gone, so I'm not confident enough
to remove it. Can Robert confirm?
> EXE_EXT - remove (unused since f06b1c598)
Daniel fixed in 88e3da565.
> endterm - remove
> (see 60c90c16c -- Use xreflabel attributes instead of endterm ...)
I read that commit message and I agree it's now unused. I just didn't
get any vibes from the commit message that it shouldn't ever be used
again. Can Tom confirm?
> xl_commit_ts_set, SizeOfCommitTsSet - remove (unused since 08aa89b32)
I would say it should be removed. I just didn't because the commit I
had pending fitted into the "typos" category and this didn't quite
fit.
I've attached a patch with the remainder.
David