Date: Wed, 7 Aug 2024 10:08:04 -0500
From: Nathan Bossart <nathandbossart@gmail.com>
To: Matthew Kim <matthewkmkim@gmail.com>
Cc: Joseph Koshakow <koshy44@gmail.com>,
	jian he <jian.universality@gmail.com>,
	Alexander Lakhin <exclusion@gmail.com>,
	Tom Lane <tgl@sss.pgh.pa.us>,
	PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
	Andres Freund <andres@anarazel.de>
Subject: Re: Remove dependence on integer wrapping
Message-ID: <ZrON1CT7Lk9_EJOG@nathan>
References: <Zp54I4WMbWW6AQnP@nathan>
 <CAAvxfHdWseDnK0gLYY51fdw-_pE=5Q2=K=pToVv5jo3YNVfHoQ@mail.gmail.com>
 <Zp7Q4erYWVjtWzQr@nathan>
 <Zp7cwd_6NXlHt6Ye@nathan>
 <CAAvxfHeg0En5VV_9pjoHgd0kerK7REL84XEiS6=PUDOSKwx9rg@mail.gmail.com>
 <CACJufxERqAgy4YBwc2hLZY=hw1kD2m6AVD3=kyAen8=HeLsfmw@mail.gmail.com>
 <CAAvxfHeOdkaQXoca3W54=RZ6opCWLzqgmDM6s8GrUxKG1Tk+_g@mail.gmail.com>
 <ZqBzEucmgeib_6Fe@nathan>
 <CAGzNdfGMe8vK1BRvUmW29T7rkqUhM67t=wbgsCSDBu=T+rqZrw@mail.gmail.com>
 <ZrKY8PJwGg23eZ3R@nathan>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <ZrKY8PJwGg23eZ3R@nathan>
Archived-At: <https://www.postgresql.org/message-id/ZrON1CT7Lk9_EJOG%40nathan>
Precedence: bulk

I started looking at 0001 again with the intent of committing it, and this
caught my eye:

-        /* make the amount positive for digit-reconstruction loop */
-        value = -value;
+        /*
+         * make the amount positive for digit-reconstruction loop, we can
+         * leave INT64_MIN unchanged
+         */
+        pg_neg_s64_overflow(value, &value);

The comment mentions that we can leave the minimum value unchanged, but it
doesn't explain why.  Can we explain why?

+static inline bool
+pg_neg_s64_overflow(int64 a, int64 *result)
+{
+    if (unlikely(a == PG_INT64_MIN))
+    {
+        return true;
+    }
+    else
+    {
+        *result = -a;
+        return false;
+    }
+}

Can we add a comment that these routines do not set "result" when true is
returned?

-- 
nathan