Re: macOS 12.6.3, static_assert, and vasnprintf compilation failure

[G. Branden Robinson]

[adding Deri]

Hi Bruno,

At 2023-02-10T21:42:09+0100, Bruno Haible wrote:
> G. Branden Robinson wrote:
> > If users of macOS of recent vintage would like to pull groff Git HEAD
> > and try a build, I'd appreciate hearing of your experiences.
> 
> The C/C++ part of the build now works fine on macOS 12.5.

Thanks for confirming this; you're the second to do so, so I feel better
about this change (dropping groff's private assert.h).

> It fails later, with
> 
>   GEN      font/devpdf/download
> Can't use string ("") as an ARRAY ref while "strict refs" in use at 
> /Users/haible/groff-1.23.0.rc2.22-55d3/build/font/devpdf/util/BuildFoundries 
> line 102, <F> line 23.
> make[2]: *** [font/devpdf/download] Error 25

Oh, bother.

Deri, we've both had our mitts on this code in the past year.  Is it
immediately obvious to you what's going on here?

As a _wild_ guess, I'm going to say that ./configure didn't find the URW
fonts.[1]  If that is true, then apparently we proceeded with an
operation we shouldn't have.  Maybe $GSpath is undefined?

87046b2948 (Werner LEMBERG      2011-07-26 17:11:12 +0000  90)  if (lc($r[0]) 
eq 'foundry')
87046b2948 (Werner LEMBERG      2011-07-26 17:11:12 +0000  91)  {
ced9cd5984 (Deri James          2022-06-21 16:38:39 +0100  92)      Warn("\nThe 
path(s) used for searching:\n".join(':',@{$foundrypath})."\n") if $notFoundFont;
87046b2948 (Werner LEMBERG      2011-07-26 17:11:12 +0000  93)      
$foundry=uc($r[1]);
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100  94)      
$foundrypath=[];
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100  95)      
push(@{$foundrypath},$dirURW) if $dirURW;
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100  96)      
push(@{$foundrypath},(split(':',$r[2])),@{$devps});
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100  97)      foreach my 
$j (0..$#{$foundrypath})
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100  98)      {
bbdc314682 (G. Branden Robinson 2022-07-13 01:53:30 -0500  99)          if 
(defined($foundrypath->[$j])
bbdc314682 (G. Branden Robinson 2022-07-13 01:53:30 -0500 100)              && 
$foundrypath->[$j]=~m'\s*\(gs\)')
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100 101)          {
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100 102)              
splice(@{$foundrypath},$j,1,@{$GSpath});
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100 103)          }
4ae4aeb655 (Deri James          2022-06-10 21:13:24 +0100 104)      }
d03f7e74fe (Deri James          2018-10-17 16:18:56 +0100 105)      
$notFoundFont=0;
87046b2948 (Werner LEMBERG      2011-07-26 17:11:12 +0000 106)  }

> Note: Rather than testers producing their own tarball, it's better if
> you provide a tarball for testing.

Agreed.  And that is normally what I do when speaking of building groff
outside of its own development mailing lists.

> The reason is: what you will release has some specific versions of
> Autoconf, Automake, Libtool, pkg.m4, and includes some specific commit
> of Gnulib. If a tester tests something different than what you will
> use in the release, there's some risk that some packaging bug will be
> unnoticed.

Yes.  I probably should have dropped bug-gnulib from the distribution or
annotated that paragraph such that it was intended for groff mailing
list readers only.

Bertrand Garrigues is the groff maintainer and he is the one entrusted
to roll and sign distribution archives.  He also handles tagging and
gnulib submodule updates.  After 1.23 is finalized, we're planning to
transition me into the maintainer role.

> So, please ask for testing on platform-testers again, when you see
> fit.

Will do.  I may ask Bertrand as soon as this weekend to roll an rc3.

Regards,
Branden

[1] https://git.savannah.gnu.org/cgit/groff.git/tree/INSTALL.extra#n131

signature.asc
Description: PGP signature