MIME-Version: 1.0
References: 
 <CAApHDvq=eEdw2Qp+aSzSOtTSe+h0fnVQ55CcTNqBkLDYiRZmxw@mail.gmail.com>
 <278B9FE3-F349-4494-99C5-483105C1C999@gmail.com>
 <CAApHDvoALVPgVSCuGdAfNUbxx_NacnAd0nuj8Hi=_-61AEqzaQ@mail.gmail.com>
 <1900289.1776212948@sss.pgh.pa.us>
 <CAApHDvq4_WDfTaWX6nHU8TV+no+aD1zjLphza685Ek7=xgGzew@mail.gmail.com>
 <1913188.1776220247@sss.pgh.pa.us>
 <CAApHDvp=fhr8ksuGyXkwYPXQpYvJ38YMhHvx6JmdD39cAvNRhw@mail.gmail.com>
 <CAApHDvrp6JwOVH3VwMB=YL3p97F5_0hKJc7J8zHRKuyhKru1dA@mail.gmail.com>
 <77f5109f-6d39-4cf1-8d0d-f826435d6cb6@app.fastmail.com>
 <CAApHDvqOgtX2MqaPmPmPET7=ySszzuoP3MLSUbWZit_V30VBLg@mail.gmail.com>
 <6cb75f47-261b-4476-b8e8-18fa6aad464f@app.fastmail.com>
In-Reply-To: <6cb75f47-261b-4476-b8e8-18fa6aad464f@app.fastmail.com>
From: David Rowley <dgrowleyml@gmail.com>
Date: Tue, 21 Apr 2026 13:40:17 +1200
Message-ID: 
 <CAApHDvqPRUqTVU0=Ewm38UKWOV5Ak1mQXg=+L98Mm9CVpyAmLQ@mail.gmail.com>
Subject: Re: Add bms_offset_members() function for bitshifting Bitmapsets
To: Greg Burd <greg@burd.me>
Cc: Tom Lane <tgl@sss.pgh.pa.us>, Chao Li <li.evan.chao@gmail.com>,
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Archived-At: 
 <https://www.postgresql.org/message-id/CAApHDvqPRUqTVU0%3DEwm38UKWOV5Ak1mQXg%3D%2BL98Mm9CVpyAmLQ%40mail.gmail.com>
Precedence: bulk

On Tue, 21 Apr 2026 at 02:55, Greg Burd <greg@burd.me> wrote:
> Were you planning on writing the optimized non-copy version as well?  I don't think it is strictly necessary, more a curiosity.
>
> bms_offset_members() -> new bms, might repalloc() replaces existing loops you've found
> bms_shift_members() -> bms is modified in place and fits your new use case a bit better

Not at this stage. The v1 patch did modify the set in-place, so the
code is there if we ever need it. I didn't find any need for it in our
current code. The selective tuple deforming patch I'm working on could
use it, but I doubt it's worth the trouble for 1 caller. It's just for
something that happens during create_plan(), so 1 more allocation in
that code likely isn't going to be noticed.

David