Horton

Re: Cannot find a working 64-bit integer type on Illumos

Thomas Munro <thomas.munro@gmail.com>

View thread View on postgr.es
From: Thomas Munro <thomas.munro@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Tomas Vondra <tomas@vondra.me>, Peter Eisentraut <peter@eisentraut.org>, Heikki Linnakangas <hlinnaka@iki.fi>, Japin Li <japinli@hotmail.com>, Andres Freund <andres@anarazel.de>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2024-12-10T02:02:07Z
Lists: pgsql-hackers

Attachments

On Thu, Dec 5, 2024 at 12:16 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Now you already snuck the camel's nose under the
> tent by including stdint.h there, and maybe these additional headers
> wouldn't do any further damage.

Even though we fixed the immediate issue (thanks), this comment stayed
with me.  I did that because I didn't want to change any interfaces at
the same time as the <stdint.h> retrofit, but I agree that it feels a
bit odd hidden in there, and doesn't appear to conform to
postgres_ext.h's self-description.  Stepping back, and I realise it's
difficult to answer with certainty, I wonder why anyone would ever
want to use postgres_ext.h directly for the definition of pg_int64
*without* being a user of libpq-fe.h.  I can't find any references to
pg_int64 (excluding forks of our code) on github; there are a few
things like proxies and suchlike that include postgres_ext.h for other
things, mostly bogusly (they also include libpq-fe.h, or they say they
want NAMEDATALEN, which isn't there anymore).

We have just three lo_*64() functions using that type and then
pg_usec_time_t.  Seems like a very narrow usage that hasn't spread,
likely only used to receive arguments, and really quite specific to
libpq-fe.h and not one of the "fundamental Postgres declarations".
Maybe we should consider moving #include <stdint.h> into libpq-fe.h?

And if we included <stdint.h> overtly, rather than covertly in
postgres_ext.h, why would we still want a third name for int64_t?  We
could change the three lo_*64() declarations to use the standard type
directly, but keep the historical typedef marked deprecated.

> But I don't see a strong argument to
> change long-standing external APIs any more than we absolutely must.

So perhaps you'll hate that idea then.  I wonder if you'll hate it
more than keeping the #include in postgres_ext.h, hence putting the
idea forward!

Commits

  1. Move pg_int64 back to postgres_ext.h

  2. pgbench: Make set_random_seed() 64-bit everywhere.

  3. Use PRI?64 instead of "ll?" in format strings (continued).

  4. Fix order of -I switches for building pg_regress.o.

  5. libpq: Deprecate pg_int64.

  6. Use PRI*64 instead of "ll*" in format strings (minimal trial)

  7. Fix header inclusion order in c.h.

  8. Use <stdint.h> and <inttypes.h> for c.h integers.

  9. Remove traces of BeOS.

  10. More correct way to check for existence of types, which allows to specify