Bug#1133510: kicad: ftbfs with GCC-16 — not reproducible with 10.0.4+dfsg-1 on amd64

Subject: Re: kicad: ftbfs with GCC-16 — not reproducible with 10.0.4+dfsg-1
on amd64

Hi,

I cannot reproduce this against 10.0.4+dfsg-1 on amd64.

I built the current unstable source in an sbuild schroot with GCC 16 and it
completed successfully:

  Toolchain package versions: binutils_2.47-2 dpkg-dev_1.23.7
  g++-16_16.1.0-3 gcc-16_16.1.0-3 libc6-dev_2.42-17
  libstdc++-16-dev_16.1.0-3 libstdc++6_16.1.0-3 linux-libc-dev_7.1.7-1

  Status: successful
  Build-Time: 23538

gcc/g++ resolved to 4:16.1.0-3 from unstable (no experimental pinning
needed, as unstable now defaults to GCC 16). Build options were
DEB_BUILD_OPTIONS="parallel=1 nocheck" with --no-arch-all, so the test
suite and the arch-all doc packages were not built; the arch-any build
itself was clean.

The failure reported against 9.0.8 was a link error in bitmap2component:

  libcommon.a(eda_base_frame.cpp.o): in function
  `KICAD_MANAGER_FRAME::GetCurrentFileName() const':
  kicad/kicad_manager_frame.h:86: undefined reference to
  `KICAD_MANAGER_FRAME::GetProjectFileName() const'

In 9.0.8, GetCurrentFileName() was defined inline in
kicad/kicad_manager_frame.h, calling GetProjectFileName(), which is only
implemented in the kicad/ application target. GCC 16 emitted an
out-of-line copy into eda_base_frame.cpp.o in libcommon, so every target
linking libcommon without the application objects failed.

In 10.0.4 that construct is gone. kicad/kicad_manager_frame.h:83 now has
only the declaration:

  wxString GetCurrentFileName() const override;

with the definition moved to the .cpp, so nothing in libcommon carries the
unresolved reference any more.

On that basis I think this can be closed, or the found version limited to
9.0.8 so that 10.0.4 is no longer marked affected and the pending
autoremoval from testing is cancelled.

One thing I have not verified: I only tested amd64.
tests.reproducible-builds.org
currently shows kicad 10.0.4+dfsg-1 as FTBFS on arm64 in both unstable and
forky, but those rebuilds ran with GCC 15, so whatever that is, it looks
like a separate issue rather than this one.

Happy to provide the full build log or re-run with different options if that
would help.