From: Tom Lane <tgl@sss.pgh.pa.us>
To: Nathan Bossart <nathandbossart@gmail.com>
cc: Thomas Munro <thomas.munro@gmail.com>,
        "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>,
        "pgsql-hackers@postgresql.org" <pgsql-hackers@postgresql.org>
Subject: Re: suppressing useless wakeups in logical/worker.c
In-reply-to: <20230126194812.GA1702315@nathanxps13>
References: <20221205173523.GA2714833@nathanxps13> <20230109174217.GA1112698@nathanxps13> <2557815.1674603908@sss.pgh.pa.us> <20230125235004.GA1327755@nathanxps13> <CA+hUKGJ=oFEX=TjeSCnGmqr_KseGLJxy=RnJMwRqPrX7N-Q2sg@mail.gmail.com> <20230126003319.GA1507251@nathanxps13> <2878152.1674700077@sss.pgh.pa.us> <CA+hUKGLpH=y4=Zx+=oLE8WvLqLHWNC-PEDARrrt=x=J_REL9Fg@mail.gmail.com> <2888102.1674705840@sss.pgh.pa.us> <3126727.1674759248@sss.pgh.pa.us> <20230126194812.GA1702315@nathanxps13>
Comments: In-reply-to Nathan Bossart <nathandbossart@gmail.com>
	message dated "Thu, 26 Jan 2023 11:48:12 -0800"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <3165442.1674763470.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Thu, 26 Jan 2023 15:04:30 -0500
Message-ID: <3165443.1674763470@sss.pgh.pa.us>
Archived-At: <https://www.postgresql.org/message-id/3165443.1674763470%40sss.pgh.pa.us>
Precedence: bulk

Nathan Bossart <nathandbossart@gmail.com> writes:
> On Thu, Jan 26, 2023 at 01:54:08PM -0500, Tom Lane wrote:
>> - * Both inputs must be ordinary finite timestamps (in current usage,
>> - * they'll be results from GetCurrentTimestamp()).
>> + * At least one input must be an ordinary finite timestamp, else the "=
diff"
>> + * calculation might overflow.  We do support stop_time =3D=3D TIMESTA=
MP_INFINITY,
>> + * which will result in INT_MAX wait time.

> I wonder if we should explicitly reject negative timestamps to eliminate
> any chance of int64 overflow, too.

Hmm.  I'm disinclined to add an assumption that the epoch is in the past,
but I take your point that the subtraction would overflow with
TIMESTAMP_INFINITY and a negative finite timestamp.  Maybe we should
make use of pg_sub_s64_overflow()?

BTW, I just noticed that the adjacent function TimestampDifference
has a lot of callers that would be much happier using
TimestampDifferenceMilliseconds.

			regards, tom lane