Hi Sébastien, Sébastien Jodogne, on 2025-08-11: > I am unable to reproduce this error on Debian unstable by > manualling compiling the package from source using the following > invocation of CMake (same invocation as in the "debian/rules" file): > > $ CC=gcc-15 CXX=g++-15 cmake ../MySQL/ [...] > > Please could someone explain me how to reproduce this issue? > > I guess this results from different versions of the dependencies > installed on my system, so I should provide the gcc/g++ version to > "pbuilder" or "gbp buildpackage", but I cannot find how to do this.
For what it's worth, I have attempted to reproduce the issue in
different configurations of sbuild (driven by gbp buildpackage),
involving Gcc 15 in sid or experimental with the below detection
by cmake
[…]
-- The C compiler identification is GNU 15.2.0
-- The CXX compiler identification is GNU 15.2.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/gcc-15 - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Check for working CXX compiler: /usr/bin/g++-15 - skipped
-- Detecting CXX compile features
-- Detecting CXX compile features - done
[…]
But I failed to reproduce any error at all. My changes to the
Salsa source tree look like
$ git diff
diff --git a/debian/control b/debian/control
index 1756079..21510bd 100644
--- a/debian/control
+++ b/debian/control
@@ -4,6 +4,8 @@ Uploaders: Sebastien Jodogne <[email protected]>
Section: science
Priority: optional
Build-Depends: cmake,
+ gcc-15,
+ g++-15,
debhelper (>= 10),
libcurl4-openssl-dev | libcurl4-dev,
libgtest-dev,
diff --git a/debian/rules b/debian/rules
index 027b9a2..540ba64 100755
--- a/debian/rules
+++ b/debian/rules
@@ -13,6 +13,9 @@ export DEB_BUILD_MAINT_OPTIONS := hardening=+all
export DEB_CFLAGS_MAINT_APPEND=-DNDEBUG
export DEB_CXXFLAGS_MAINT_APPEND=-DNDEBUG
+export CC = gcc-15
+export CXX = g++-15
+
%:
dh $@ --builddirectory=Build
I have also tested a variant involving gcc 15 as default version
of gcc in experimental without much change in the behavior of
the build. I have uploaded my full build log of the sid build
variant on the people's web server[1] to facilitate pinpointing
the relevant change. My current conjecture is, that a change in
Boost could have removed a <cassert> import. There is a wave of
bugs open about the upcoming Boost 1.88 and it's possible that
the entry missed that crucial information.
[1]:
https://people.debian.org/~emollier/logs/orthanc-mysql/orthanc-mysql_5.0+dfsg-1_amd64-gcc-15.build.xz
In hope this helps,
--
.''`. Étienne Mollier <[email protected]>
: :' : pgp: 8f91 b227 c7d6 f2b1 948c 8236 793c f67e 8f0d 11da
`. `' sent from /dev/pts/2, please excuse my verbosity
`- on air: The Light Year - Steps to Life
signature.asc
Description: PGP signature
