Horton

Thread

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Fix sign-compare warnings in pg_iovec.h.

  1. Fix port/pg_iovec.h building extensions on x86_64-darwin

    Wolfgang Walther <walther@technowledgy.de> — 2024-11-08T19:08:06Z

    When building pg_cron [1] or pg_hll [2] for PG 17 on x86_64-darwin, we 
encounter the following build failure in nixpkgs:


/nix/store/2clw9wg933c40f871d8iqbl909lg2yis-postgresql-17.0-dev/include/server/port/pg_iovec.h:71:12: 
error: comparison of integers of different signs: 'ssize_t' (aka 'long') 
and 'const size_t' (aka 'const unsigned long') [-Werror,-Wsign-compare]
                 if (part < iov[i].iov_len)
                     ~~~~ ^ ~~~~~~~~~~~~~~
/nix/store/2clw9wg933c40f871d8iqbl909lg2yis-postgresql-17.0-dev/include/server/port/pg_iovec.h:110:12: 
error: comparison of integers of different signs: 'ssize_t' (aka 'long') 
and 'const size_t' (aka 'const unsigned long') [-Werror,-Wsign-compare]
                 if (part < iov[i].iov_len)
                     ~~~~ ^ ~~~~~~~~~~~~~~
2 errors generated.


The attached patch fixes those.

Hopefully this can make it into the minor release next week.

Best,

Wolfgang

[1]: https://hydra.nixos.org/build/276421287/nixlog/1
[2]: https://hydra.nixos.org/build/276419879/nixlog/1

  2. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Nathan Bossart <nathandbossart@gmail.com> — 2024-11-08T20:00:36Z

    On Fri, Nov 08, 2024 at 08:08:06PM +0100, Wolfgang Walther wrote:
> @@ -68,7 +68,7 @@ pg_preadv(int fd, const struct iovec *iov, int iovcnt, off_t offset)
>  		}
>  		sum += part;
>  		offset += part;
> -		if (part < iov[i].iov_len)
> +		if ((size_t) part < iov[i].iov_len)
>  			return sum;
>  	}
>  	return sum;
> @@ -107,7 +107,7 @@ pg_pwritev(int fd, const struct iovec *iov, int iovcnt, off_t offset)
>  		}
>  		sum += part;
>  		offset += part;
> -		if (part < iov[i].iov_len)
> +		if ((size_t) part < iov[i].iov_len)
>  			return sum;
>  	}
>  	return sum;

This looks correct to me.  At this point in the code, we know that part >=
0, so casting it to an unsigned long ought to be okay.

-- 
nathan

  3. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Nathan Bossart <nathandbossart@gmail.com> — 2024-11-08T21:20:00Z

    Here is a new version of the patch with an updated commit message.  The
cfbot results [0] look good, so I plan to commit this one shortly unless
someone objects.

[0] https://cirrus-ci.com/build/6599760059039744

-- 
nathan

  4. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Nathan Bossart <nathandbossart@gmail.com> — 2024-11-08T22:15:47Z

    On Fri, Nov 08, 2024 at 03:20:00PM -0600, Nathan Bossart wrote:
> Here is a new version of the patch with an updated commit message.  The
> cfbot results [0] look good, so I plan to commit this one shortly unless
> someone objects.

Committed.

-- 
nathan

  5. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Thomas Munro <thomas.munro@gmail.com> — 2024-11-08T22:26:44Z

    On Sat, Nov 9, 2024 at 8:08 AM Wolfgang Walther <walther@technowledgy.de> wrote:
> When building pg_cron [1] or pg_hll [2] for PG 17 on x86_64-darwin, we
> encounter the following build failure in nixpkgs:

Out of curiosity, is nixos deliberately using an old macOS deployment
target or SDK, 10.something?  I'm just wondering if our feature
detection is working correctly on macOS/x86, because I'd expect real
preadv/pwritev to be there from 11 onwards, and 11 is already out of
support by Apple.

  6. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Nathan Bossart <nathandbossart@gmail.com> — 2024-11-08T22:36:18Z

    On Sat, Nov 09, 2024 at 11:26:44AM +1300, Thomas Munro wrote:
> Out of curiosity, is nixos deliberately using an old macOS deployment
> target or SDK, 10.something?  I'm just wondering if our feature
> detection is working correctly on macOS/x86, because I'd expect real
> preadv/pwritev to be there from 11 onwards, and 11 is already out of
> support by Apple.

It appears to be working for longfin [0].

[0] https://buildfarm.postgresql.org/cgi-bin/show_stage_log.pl?nm=longfin&dt=2024-11-08%2021%3A36%3A02&stg=configure

-- 
nathan

  7. Re: Fix port/pg_iovec.h building extensions on x86_64-darwin

    Wolfgang Walther <walther@technowledgy.de> — 2024-11-09T11:06:07Z

    Thomas Munro:
> Out of curiosity, is nixos deliberately using an old macOS deployment
> target or SDK, 10.something?  I'm just wondering if our feature
> detection is working correctly on macOS/x86, because I'd expect real
> preadv/pwritev to be there from 11 onwards, and 11 is already out of
> support by Apple.

There has been a lot of work on the macOS SDK in nixpkgs lately, but it 
seems that the default for x86-64 is still v10 [1], yes:

 > x86_64-darwin uses the 10.12 SDK by default, but some software is not
 > compatible with that version of the SDK. In that case, the 11.0 SDK
 > used by aarch64-darwin is available for use on x86_64-darwin. [...]

It seems like the goal is to bump x86-64 to v11 in the next release 
cycle [2].

Best,

Wolfgang

[1]: https://nixos.org/manual/nixpkgs/unstable/#sec-darwin
[2]: 
https://discourse.nixos.org/t/on-the-future-of-darwin-sdks-or-how-you-can-stop-worrying-and-put-the-sdk-in-build-inputs/50574/11