Bug 928096 – app-text/texlive-core-2023-r10 fails to compile: pdftexini.c:4015:25: error: assignment to boolean {aka int } from incompatible pointer type internalfontnumber {aka long int } [-Wincompatible-pointer-types] (32bit arches)

• Status: CONFIRMED
• Product: Gentoo Linux

[Description Agostino Sarubbo 2024-03-29 07:53:42 UTC]

https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/

Issue: app-text/texlive-core-2023-r5 fails to compile.
Discovered on: x86 (internal ref: tinderbox_x86)
System: GCC-14-SYSTEM (https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#GCC-14)

Info about the issue:
https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#CF0003

[Comment 1 Agostino Sarubbo 2024-03-29 07:53:43 UTC]

Created attachment 888969 [details]
build.log.xz

build log and emerge --info (compressed because it exceeds attachment limit, use 'xzless' to read it)

[Comment 2 Florian Schmaus 2024-04-04 07:22:44 UTC]

pdftexini.c: In function 'loadfmtfile':
pdftexini.c:4015:25: error: assignment to 'boolean *' {aka 'int *'} from incompatible pointer type 'internalfontnumber *' {aka 'long int *'} [-Wincompatible-pointer-types[https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#index-Wincompatible-pointer-types]]
 4015 |     pdffonthasspacechar = xmallocarray ( internalfontnumber , fontmax ) ;
      |                         ^
pdftexini.c: In function 'mainbody':
pdftexini.c:5568:27: error: assignment to 'boolean *' {aka 'int *'} from incompatible pointer type 'internalfontnumber *' {aka 'long int *'} [-Wincompatible-pointer-types[https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#index-Wincompatible-pointer-types]]
 5568 |       pdffonthasspacechar = xmallocarray ( internalfontnumber , fontmax ) ;
      |                           ^
make[5]: *** [Makefile:17857: pdftex-pdftexini.o] Error 1

[Comment 3 Florian Schmaus 2024-04-04 12:41:10 UTC]

It seems that pdftexini.c is not part of the source code of texlive, but rather somehow generated? Then again, I am not yet able to locate the part where it is generated.

[Comment 4 Agostino Sarubbo 2024-09-22 09:48:22 UTC]

tinderbox_x86 has reproduced this issue with version 2023-r10 - Updating summary.

[Comment 5 Andreas K. Hüttel 2024-11-19 22:53:07 UTC]

(In reply to Florian Schmaus from comment #3)
> It seems that pdftexini.c is not part of the source code of texlive, but
> rather somehow generated? Then again, I am not yet able to locate the part
> where it is generated.

web2c-sh

[Comment 6 Andreas K. Hüttel 2024-11-19 22:57:50 UTC]

Can't reproduce this on amd64. It may be that the number sizes are arch-dependent.

[Comment 7 Andreas K. Hüttel 2024-11-20 11:50:47 UTC]

Confirmed on x86

pdftexini.c: In function 'loadfmtfile':
pdftexini.c:4015:25: error: assignment to 'boolean *' {aka 'int *'} from incompatible pointer type 'internalfontnumber *' {aka 'long int *'} [-Wincompatible-pointer-types]
 4015 |     pdffonthasspacechar = xmallocarray ( internalfontnumber , fontmax ) ;
      |                         ^
pdftexini.c: In function 'mainbody':
pdftexini.c:5568:27: error: assignment to 'boolean *' {aka 'int *'} from incompatible pointer type 'internalfontnumber *' {aka 'long int *'} [-Wincompatible-pointer-types]
 5568 |       pdffonthasspacechar = xmallocarray ( internalfontnumber , fontmax ) ;
      |                           ^
make[5]: *** [Makefile:17888: pdftex-pdftexini.o] Error 1

[Comment 8 Larry the Git Cow 2024-11-20 12:49:35 UTC]

The bug has been referenced in the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=537d1d2023912f6115f4fce9aa3937d8966e76d9

commit 537d1d2023912f6115f4fce9aa3937d8966e76d9
Author:     Andreas K. Hüttel <[e-mail]>
AuthorDate: 2024-11-20 12:47:24 +0000
Commit:     Andreas K. Hüttel <[e-mail]>
CommitDate: 2024-11-20 12:49:14 +0000

    app-text/texlive-core: gcc-14 build workaround for 32bit arches
    
    This is upstream recommendation for the moment, see also
    https://www.tug.org/texlive/build.html
    I'm fairly sure it just hides a real bug in pdftex, keeping 928096
    thus open, but hey, at least it's not a regression...
    
    Bug: https://bugs.gentoo.org/928096
    Signed-off-by: Andreas K. Hüttel <[e-mail]>

 app-text/texlive-core/texlive-core-2023-r10.ebuild | 7 +++++++
 1 file changed, 7 insertions(+)