Opened 17 months ago
Closed 16 months ago
#21500 closed enhancement (fixed)
Fixes for packages that FTBS with gcc-15
| Reported by: | Joe Locash | Owned by: | blfs-book |
|---|---|---|---|
| Priority: | normal | Milestone: | 12.4 |
| Component: | BOOK | Version: | git |
| Severity: | medium | Keywords: | |
| Cc: |
Description (last modified by )
Leaving open for now, but see https://wiki.linuxfromscratch.org/blfs/wiki/gcc15-todo for details.
The items below are out of date. ============
cpio
https://savannah.gnu.org/bugs/?66297Patch implemented as a sed in [90e2079c1a]
- cyrus-sasl-2.1.28
https://github.com/cyrusimap/cyrus-sasl/pull/873
Note: upstream rejected it because the patch doesn't apply to 2.2 that is in the master branch (that code no longer exists)
- lsof-4.99.0
- parted-3.6
https://git.savannah.gnu.org/cgit/parted.git/commit/?id=16343bda6ce0d41edf43f8dac368db3bbb63d271
- sendmail-8.18.1
https://src.fedoraproject.org/rpms/sendmail/raw/rawhide/f/sendmail-8.18.1-gcc-15-fix.patch
- libburn-1.5.6
- cifs-utils-7.3
run autoreconf -fi before configure
- freeglut-3.6.0
https://github.com/freeglut/freeglut/commit/800772e993a3ceffa01ccf3fca449d3279cde338
- mupdf-1.25.6
https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=dbb84af6533b0c940d1973199fbe6d84e25589d1
Note: this patch just fixes thirdparty/freeglut
- xvidcore-1.3.7
sed -i '/^typedef int bool;/d' src/encoder.h
- cups-filters-2.0.1
https://github.com/OpenPrinting/cups-filters/commit/9871a50b5c1f9c2caa2754aac1f5db70c886021b
Attachments (2)
Change History (22)
comment:1 by , 17 months ago
| Description: | modified (diff) |
|---|
follow-ups: 8 11 comment:5 by , 17 months ago
Replying to Xi Ruoyao:
Can we use system freegult for mupdf instead?
I tested and it builds fine using the system freeglut.
comment:6 by , 17 months ago
- procmail-3.24
The last commit to this project was 3 years ago to release 3.24. There are 2 pull requests still pending to fix building with gcc-14 and clang. An issue (https://github.com/BuGlessRB/procmail/issues/10) about building with gcc-15 was opened on Feb 28 with no replies. This looks abandoned to me. The book currently patches for gcc-14.
I propose to drop the patch and just build with:
make CC="gcc -std=gnu89" LOCKINGTEST=/tmp MANDIR=/usr/share/man install
comment:7 by , 17 months ago
Downgrading the standard should be only the last resort if no other options are available. The code pattern triggering the error in the new standard still invokes undefined behavior in the old standard.
comment:8 by , 17 months ago
Replying to Joe Locash:
Replying to Xi Ruoyao:
Can we use system freegult for mupdf instead?
I tested and it builds fine using the system freeglut.
See #21506.
comment:9 by , 17 months ago
- ghostscript-10.05.1
Upstream fix: https://cgit.ghostscript.com/cgi-bin/cgit.cgi/ghostpdl.git/commit/?id=ae940946473ceb8c5353bc6e7f04673c6e60502d
I'll attach a refactored patch for it to build with 10.05.1.
In the commit it patches 3 files which are not in the 10.05.x release: pcl/pcl/pcpage.h, pcl/pl/plmain.h, and xps/xpstop.c. It also patches base/gxccache.c but that very small snippet of code doesn't exist in 10.05.x. The other change I made was changing use_padding from an int to a bool in psi/zfaes.c to match the prototype of dict_bool_param().
by , 17 months ago
| Attachment: | ghostscript-10.05.1-gcc15.patch added |
|---|
comment:10 by , 17 months ago
Thanks for your good work on this Joe. We really needed to wait for upstream to do this because of how invasive it was.
comment:11 by , 17 months ago
Replying to Joe Locash:
Replying to Xi Ruoyao:
Can we use system freegult for mupdf instead?
I tested and it builds fine using the system freeglut.
Fixed at 14b753e29d999cd09b1aa6fd059cf311b8e9da56 (BLFS Ticket #21506)
follow-up: 16 comment:12 by , 17 months ago
gimp:
sed -ri 's/bool([; ])/boolval\1/' libgimpconfig/gimpconfig-serialize.c
sed -i 's/enum { false = 0, true = 1 }/bool/' plug-ins/selection-to-path/types.h
follow-up: 14 comment:13 by , 17 months ago
Firefox:
As time of writing, it was version 128.10.1.
There is a bug report at https://bugzilla.mozilla.org/show_bug.cgi?id=1917964 closed 8 month ago but neither the small patch nor the big one (referred to in that ticket) has made it into the 128 series.
A small sed fixes the issue:
sed -e "s/<mask_type, N>/<mask_index, N>/" \ -i gfx/wr/swgl/src/vector_type.h
follow-up: 15 comment:14 by , 17 months ago
Replying to thomas:
Firefox:
As time of writing, it was version 128.10.1.
There is a bug report at https://bugzilla.mozilla.org/show_bug.cgi?id=1917964 closed 8 month ago but neither the small patch nor the big one (referred to in that ticket) has made it into the 128 series.
A small sed fixes the issue:
sed -e "s/<mask_type, N>/<mask_index, N>/" \ -i gfx/wr/swgl/src/vector_type.h
No, we build firefox with clang.
comment:15 by , 17 months ago
Replying to Xi Ruoyao:
Replying to thomas:
Firefox:
As time of writing, it was version 128.10.1.
There is a bug report at https://bugzilla.mozilla.org/show_bug.cgi?id=1917964 closed 8 month ago but neither the small patch nor the big one (referred to in that ticket) has made it into the 128 series.
A small sed fixes the issue:
sed -e "s/<mask_type, N>/<mask_index, N>/" \ -i gfx/wr/swgl/src/vector_type.hNo, we build firefox with clang.
No, not 'No' - it is an issue. Book still mentions the option to use gcc. Btw, the statement in book that building for i686 is broken is false (as long as the sse cflags are set).
comment:16 by , 17 months ago
Replying to Xi Ruoyao:
gimp:
sed -ri 's/bool([; ])/boolval\1/' libgimpconfig/gimpconfig-serialize.c sed -i 's/enum { false = 0, true = 1 }/bool/' plug-ins/selection-to-path/types.h
Version 3.0.4 compiles fine with gcc15 without any tweaks.
comment:17 by , 16 months ago
| Description: | modified (diff) |
|---|
comment:18 by , 16 months ago
| Resolution: | → duplicate |
|---|---|
| Status: | new → closed |
comment:19 by , 16 months ago
| Description: | modified (diff) |
|---|---|
| Resolution: | duplicate |
| Status: | closed → reopened |
comment:20 by , 16 months ago
| Resolution: | → fixed |
|---|---|
| Status: | reopened → closed |
I think we've got everything mentioned in this ticket resolved. If something new comes up or something we did here needs a revision, then let's open a new ticket for that packages.
Marking fixed.
