Hi,

During a rebuild of all packages in sid, your package failed to build
on amd64.

Relevant part (hopefully):
http://qa-logs.debian.net/2020/12/26/evolution-data-server_3.38.2-2_unstable.log

A list of current common problems and possible solutions is available at
http://wiki.debian.org/qa.debian.org/FTBFS . You're welcome to contribute!

If you reassign this bug to another package, please marking it as 'affects'-ing
this package. See https://www.debian.org/Bugs/server-control#affects

If you fail to reproduce this, please provide a build log and diff it with me
so that we can identify if something relevant changed in the meantime.

About the archive rebuild: The rebuild was done on EC2 VM instances from
Amazon Web Services, using a clean, minimal and up-to-date chroot. Every
failed build was retried once to eliminate random failures.

Hello,

Seems like intentional(?) changes in libphonennumber caused this.

Adding recipient Matthias Klose, who uploaded the latest libphonenumber
version, to hopefully get a comment on if this is something than should
get fixed in libphonenumber or in e-d-s.

Please note that the libphonenumber Vcs-Git repo seems outdated so it's
not possible to track down exact commit which changed debian/control and
get more details!

Nope, this is a wild goose....
[...]


Hopefully actually relevant part (because I have to say finding the
actual error from CMake builds is horrible!):

```
gmake[3]: Entering directory '/<<PKGBUILDDIR>>/obj-x86_64-linux-gnu/CMakeFiles/CMakeTmp'
Building CXX object CMakeFiles/cmTC_24d88.dir/src.cxx.o
/usr/bin/c++ -DI18N_PHONENUMBERS_USE_BOOST  -g -O2 -ffile-prefix-map=/<<PKGBUILDDIR>>=. -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -Dphone_number_with_boost_thread-mt -fPIE -o CMakeFiles/cmTC_24d88.dir/src.cxx.o -c /<<PKGBUILDDIR>>/obj-x86_64-linux-gnu/CMakeFiles/CMakeTmp/src.cxx
<command-line>: warning: ISO C++11 requires whitespace after the macro name
In file included from /usr/include/phonenumbers/phonenumberutil.h:32,
                 from /<<PKGBUILDDIR>>/obj-x86_64-linux-gnu/CMakeFiles/CMakeTmp/src.cxx:1:
/usr/include/phonenumbers/base/memory/scoped_ptr.h:10:10: fatal error: boost/scoped_ptr.hpp: No such file or directory
   10 | #include <boost/scoped_ptr.hpp>
      |          ^~~~~~~~~~~~~~~~~~~~~~
compilation terminated.
gmake[3]: *** [CMakeFiles/cmTC_24d88.dir/build.make:85: CMakeFiles/cmTC_24d88.dir/src.cxx.o] Error 1
gmake[3]: Leaving directory '/<<PKGBUILDDIR>>/obj-x86_64-linux-gnu/CMakeFiles/CMakeTmp'
gmake[2]: *** [Makefile:140: cmTC_24d88/fast] Error 2
```

So this looks like a bug in libphonenumber to me.... Interestingly
this is part of libphonenumber 7.1.0-7 changelog:
 * Don't build using boost, not required on recent Linux versions.

... and in /usr/include/phonenumbers/base/memory/scoped_ptr.h we can
see that the boost header is conditionally included based on
`#if defined(I18N_PHONENUMBERS_USE_BOOST)`.

That define in turn is apparently set by evolution-data-server:
./cmake/modules/FindPhonenumber.cmake:set(PHONENUMBER_DEFINITIONS -DI18N_PHONENUMBERS_USE_BOOST CACHE STRING "libphonenumber compile definitions, default is -DI18N_PHONENUMBERS_USE_BOOST")

Does that mean evolution-data-server is supposed to open-code the
dependencies needed by libphonenumber on boost ?
I think it would be better if the libphonenumber-dev pulled in
everything needed in every supported configuration (or alternatively
provided 2 separate -dev packages, eg. libphonenumber-boost-dev which
deps on libphonennumber-dev plus required boost parts but that seems
too complex for little gain - not to mention needing to go through NEW).



Regards,
Andreas Henriksson

severity 978172 normal
retitle 978172 Decide whether libphonenumber or its users should B-D
against libboost dev packages
thanks

EDS builds fine again, now that libphonenumber has reintroduced the B-D
against libboost dev packages

libphonenumber maintainer wants use to depend on boost dev package,
let's discuss that after bullseye release

Hi Laurent, Matthias, Andreas:

I am reviving the discussion now that bullseye has been released.  The
commit that added the B-D on libboost-thread-dev to libphonenumber-dev
is [1].

What are the drawbacks to leaving the boost B-D in libphonenumber-dev
and closing the bug?

Thank you,
tony

[1] https://salsa.debian.org/debian/libphonenumber/-/commit/be4e4215dcf0260344499ac05d9199ff55748cb7

[ adding #978172 to Cc ]

Hi Tony,

yes, this is a similar issue, and I do also see the dependency as
mandatory there:

When libphonenumber8 is only using a library without exposing it in its
API, then there is no need for a dependency in libphonenumber-dev.[1]

When a library is exposing its own dependencies in its API,
by #include'ing of headers of the libraries it uses, then
the dependencies should provide everything necessary for
using the headers.

cu
Adrian

[1] Ignoring the static libraries in libphonenumber-dev, in practice
    static libraries are rarely used and -dev packages do not usually
    have dependencies for using them.