Re: (gnuradio-core) libstdc++.so.9 conflicts with libstdc++.so.7

[Havard Eidnes <he%NetBSD.org@localhost>]

>> http://www.ki.nu/pkgsrc/reports/current/NetBSD-9.0/20221116.2248/gnuradio-core-3.10.4.0nb1/build.log
>> +--------------------------------------------------------------------
>> | [ 53%] Linking CXX shared library libgnuradio-blocks.so /usr/bin/ld:
>> | warning: libstdc++.so.9, needed by
>> | /tmp/ham/gnuradio-core/work/.buildlink/lib/libboost_program_options.so,
>> | may conflict with libstdc++.so.7
>> | 
>> | [ 53%] Built target gnuradio-blocks
>> | gmake: *** [Makefile:146: all] Error 2
>> +---------------------------------------------------------------------
>> (I don't see the origin of  libstdc++.so.7)
>
> This is usually caused by some dependency setting GCC_REQD.

Right.  This may be a variation over the theme I talked about
earlier under the "libstdc++.so and mysterious build errors"
subject, although in that case it was an old libstdc++.so.7 left
in /usr/lib which "should" have been picked up from the gcc10
pacage's lib directory instead which was the issue.  But this
will then be the next problem hitting once that's resolved...

I'm guessing also here the fact that pkgsrc gcc and the base
system gcc are not aligned with respect to the major version
number for libstdc++ is a major part of this problem.  NetBSD-
current uses gcc version 10 with libstdc++.so.9 while pkgsrc
gcc10 uses libstdc++.so.7 for the "same" if I am not terribly
mistaken.  I'm not sure it's a good idea to try to mix the use of
both of those in one and the same process (this may be an
understatement).  However, the cure to this problem as things
stands now is far from straight-forward, unfortunately (one way
could be to basically "force pkgsrc gcc10 for everything except,
of course, for the gcc10 package itself, and the things it
depends on...").

Another way to get exposed to this is to require the use of
Fortran (e.g. the various R packages), which forces the use of
pkgsrc gcc(10?).

Regards,

- Håvard