block 1128550 by 1124418 thanks Sorry about leaving gnulib in this state, the intent was indeed to upload to experimental, but somehow my 'gbp dch' spell misfired. I have been waiting on the 'git-merge-changelog' ITP to be approved, and I asked the DFSG team to consider reviewing it sooner.
To clarify, did you confirm that gnulib from latest upstream 'stable-202601' solved the FTBFS in inetutils? If so I'll try to backport the change on that branch into a new unstable 'gnulib' upload. The Debian gnulib package uses upstream's gnulib git bundle as source code, which doesn't have this patch in it yet, but I suppose it could go into debian/patches/ meanwhile. It won't still migrate to testing until the git-merge-changelog situation is resolved, I think, but I suppose that is not as urgent. Getting gnulib into testing is blocking on that ITP, so I added a blocker. Adding a Breaks seems a bit blunt: there are many valid use-cases of current gnulib that won't Break with recent libc6-dev. But adding it doesn't seem to violate any policy, so no strong opinion from me. I'm just not sure if adding it actually leads to an overall improvement. /Simon Guillem Jover <[email protected]> writes: > Control: reassign -1 gnulib > Control: affects -1 inetutils > > Hi! > > On Fri, 2026-02-20 at 23:58:01 +0100, Aurelien Jarno wrote: >> Source: inetutils >> Version: 2:2.7-3 >> Severity: important >> Tags: ftbfs upstream >> Justification: fails to build from source >> User: [email protected] >> Usertags: glibc-2.43 > >> inetutils fails to build from source with glibc 2.43, currently in >> experimental. From the build log: >> >> | gcc -DHAVE_CONFIG_H -I. -I.. -Wdate-time -D_FORTIFY_SOURCE=2 >> -Wno-cast-qual -Wno-conversion -Wno-float-equal -Wno-sign-compare -Wno-undef >> -Wno-unused-function -Wno-unused-parameter -Wno-float-conversion >> -Wimplicit-fallthrough -Wno-pedantic -Wno-sign-conversion -Wno-type-limits >> -Wno-unused-const-variable -Wno-unsuffixed-float-constants -Wno-error -Wall >> -g -O2 -Werror=implicit-function-declaration >> -ffile-prefix-map=/build/reproducible-path/inetutils-2.7=. >> -fstack-protector-strong -fstack-clash-protection -Wformat >> -Werror=format-security -mbranch-protection=standard -c -o >> libgnu_a-c32tolower.o `test -f 'c32tolower.c' || echo './'`c32tolower.c >> | In file included from /usr/include/features.h:539, >> | from >> /usr/include/aarch64-linux-gnu/bits/libc-header-start.h:33, >> | from /usr/include/stdlib.h:26, >> | from ./stdlib.h:51, >> | from argp-fmtstream.c:26: >> | ./stdlib.h:820:20: error: expected identifier or '(' before '_Generic' >> | 820 | _GL_EXTERN_C void *bsearch (const void *__key, >> | | ^~~~~~~ > […] >> | dh_auto_build: error: make -j128 returned exit code 2 >> | make: *** [debian/rules:29: binary] Error 25 >> | dpkg-buildpackage: error: debian/rules binary subprocess failed >> | with exit status 2 >> >> The full build log is available here [1]. >> >> The issue is due to ISO C23 declaration of bsearch, memchr, strchr, >> strpbrk, strrchr, strstr, wcschr, wcspbrk, wcsrchr, wcsstr and wmemchr, >> which now returns a pointer to a const-qualified type when the input >> argument is a pointer to a const-qualified type [2]. > > The problematic code is in gnulib, which inetutils pulls in at build > time. This has been fixed in upstream gnulib at least with commit > df17f4f37ed3ca373d23ad42eae51122bdb96626 (there might be other commits > needed besides that though). > > There was a gnulib upload with a recent version that I think should > contain the fixed code, which looks being intended for experimental, > but ended up in unstable, and even though the buildd status says it > supposedly got built, it does not appear on unstable. > > > https://tracker.debian.org/news/1706111/accepted-gnulib-20260109-1exp1-source-into-unstable/ > https://buildd.debian.org/status/package.php?p=gnulib > > I tested with a local checkout (from today) of gnulib and the glibc > from experimental, and inetutils then builds fine. So a fixed gnulib > should fix the inetutils FTBFS. > > Once a fixed gnulib is uploaded, it might also make sense to add a > version Breaks from libc6-dev against gnulib probably. > > Thanks, > Guillem
signature.asc
Description: PGP signature
