Re: Patch for bind message to clarify signedness of parameters and result column format codes
Dave Cramer <davecramer@gmail.com>
From: Dave Cramer <davecramer@gmail.com>
To: Peter Eisentraut <peter@eisentraut.org>
Cc: Pg Docs <pgsql-docs@lists.postgresql.org>
Date: 2026-06-10T14:03:40Z
Lists: pgsql-docs
Attachments
- 0002-doc-clarify-signedness-of-integer-fields-in-protocol.patch (application/octet-stream) patch 0002
On Wed, 10 Jun 2026 at 09:16, Peter Eisentraut <peter@eisentraut.org> wrote: > On 09.06.26 13:51, Dave Cramer wrote: > > Currently we don't say whether the number of parameters is signed or > > unsigned. Same with results > > Instead of repeating this with every message, we should clarify this in > the section "Message Data Types". > I've attached a new patch. > > I had always assumed that a type designation like Int16 or Int32 is > signed, because of how they look like C types. But on the other hand, > the POSIX functions to convert between host and network order only deal > with unsigned types. So it's not obvious either way. > > It was surprising to me that some of them were unsigned Dave