Robert Haas <robertmhaas@gmail.com> writes:
> On Sun, Jun 21, 2020 at 7:22 AM Tomas Vondra
> <tomas.vondra@2ndquadrant.com> wrote:
>> The reason why I kept the single-word variant is consistency with other
>> GUCs that affect planning, like enable_indexscan, enable_hashjoin and
>> many others.
> Right, so that makes sense, but from a larger point of view, how much
> sense does it actually make?
Maybe I'm just used to the names, but I find that things like
"enable_seqscan" and "enable_nestloop" are pretty readable.
Once they get longer, though, not so much. So I agree with
renaming enable_incrementalsort.
> So I'm +1 for changing this, and I'd definitely be +1 for renaming the
> others if they weren't released already, and at least +0.5 for it
> anyhow.
Nah. Those names are way too well entrenched. Besides which, if
we open them up for reconsideration, there's going to be a lot of
bikeshedding done. Should "enable_seqscan" become "enable_seq_scan",
or "enable_sequential_scan", or maybe "enable_scan_sequential"?
Why doesn't "enable_nestloop" contain the word "join"? Etc etc.
(I do have to wonder if maybe this one should be enable_sort_incremental.)
regards, tom lane