Getting rid of aggregate_dummy()

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: pgsql-hackers(at)lists(dot)postgresql(dot)org
Subject: Getting rid of aggregate_dummy()
Date: 2020-11-01 20:47:45
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers

While fooling with for a nearby patch [1], I noticed
that fmgrtab.c had a lot of entries pointing at aggregate_dummy,
which seemed rather useless. So I experimented with removing them.

It turns out that nodeWindowAgg.c is carelessly expecting them to be
there, because it does fmgr_info_cxt() on the target window function
even if it will never call it because it's a plain aggregate.
But that's pretty trivial to fix, just need to relocate that call.

With that, we don't actually need aggregate_dummy() to exist at
all, because it's never referenced. Having "aggregate_dummy"
as the prosrc value for an aggregate function is now just a
random convention; any other string would do as well. (We could
save a few bytes in pg_proc by choosing a shorter string, but
probably it's better to stick to the existing convention.)

Anyway, this saves about 3KB in fmgrtab.o, without any downside
that I can see. If someone accidentally called an aggregate as
a normal function, they'd now get a different error message,
namely "internal function "aggregate_dummy" is not in internal lookup
table" instead of "aggregate function NNN called as normal function".
That doesn't really seem like a problem.

The attached patch is a delta over the one in [1].

regards, tom lane


Attachment Content-Type Size
remove-aggregate_dummy.patch text/x-diff 4.0 KB


Browse pgsql-hackers by date

  From Date Subject
Next Message Daniel Gustafsson 2020-11-01 22:04:18 Re: Support for NSS as a libpq TLS backend
Previous Message Tom Lane 2020-11-01 19:19:44 Getting to generate macros for all pg_proc entries