Bugs or pull requests that clearly needs to be closed, merged or rejected):
Should be backported to maint-3.9
Mmm, this one seems like a feature creep and something not generally useful. If we he really wants to postprocess data in excel he might as well just dump it to a file and read that into excel.
Followup on:
gnuradio/gnuradio#4032 (yayayayay)
https://github.com/gnuradio/gnuradio/blob/master/cmake/Modules/GrSwig.cmake#L115-L126
[20:55:38] <nickoe> http://dpaste.com/21A4BDK.txt
[20:55:55] <nickoe> master
[20:56:13] <nickoe> I just tried a clean build thinking I had some cmakecache artefacts
[21:07:30] <slack-gw5> <indianam1ke> @nickoe : is there a GR version already installed, maybe in CMAKE_INSTALL_PREFIX ?
[21:08:03] <nickoe> I have the system gnuradio installed, but the destimation for the install prefix is not there yet
[21:08:21] <nickoe> I mean, I did rm -rf it before
[21:08:41] <nickoe> I usually install to /tmp/gnuradio
[21:09:15] <slack-gw5> <indianam1ke> Current GR doesn’t build cleanly if a prior GR install is located in standard system paths for headers
[21:09:56] <nickoe> mmm, that is wonky
[21:10:08] <nickoe> How can that be?
[21:10:26] <slack-gw5> <indianam1ke> yes . it is … quite a PITA for those of us who do those sorts of builds!
[21:11:01] <slack-gw5> <indianam1ke> It’s an artifact of the way CMake internally orders `include_directories` as set via various methods
[21:11:03] <nickoe> I don't see it reffering to /usr/include or anything
[21:11:34] <nickoe> That certainly sounds like a bug to me
[21:11:55] <slack-gw5> <indianam1ke> can you find the actual compile command for `digital_swig2PYTHON_wrap.cxx` in the log?
[21:12:10] <slack-gw5> <indianam1ke> it’s a design decision (I think and hope) for how CMake works
[21:12:33] <nickoe> I am sure we are doing something wrong with cmake
[21:13:57] <nickoe> https://github.com/gnuradio/gnuradio/issues/3054
[21:14:09] <slack-gw5> <indianam1ke> I investigated this a while back … it comes down to what I note … I could not find a _robust_ way around the issue … there were some hacky ways, and they were horrible
[21:14:26] <nickoe> ... https://github.com/gnuradio/gnuradio/issues/3054#issuecomment-573091413
[21:14:36] <nickoe> it is certainly unreasonable to require installing stuff
[21:15:00] <slack-gw5> <indianam1ke> yup … I think we (GR Leads) discussed this a while back & decided to just wait for PyBIND11 …
[21:15:10] <slack-gw5> <indianam1ke> this doesn’t help GR38 or prior, of course
[21:15:42] <nickoe> but what cmake file does the problem to this stem from?
[21:15:57] <slack-gw5> <indianam1ke> not one specific file unfortunately!
[21:16:03] <nickoe> but I am on master
[21:16:11] <nickoe> isn't that sorta considered 3.9?
[21:16:28] <slack-gw5> <indianam1ke> yes … 3.9-berta sort of
[21:17:39] <slack-gw5> <indianam1ke> if you want to try debugging, here’s where to target (_finding a link … gimme a sec)_
[21:18:52] <slack-gw5> <indianam1ke> https://github.com/gnuradio/gnuradio/blob/master/cmake/Modules/GrSwig.cmake#L115-L126
[21:19:21] <slack-gw5> <indianam1ke> Add debugging printouts to those lines of code to see what the actual `include_directories` end up being
Unuseful bugs:
--
Old bugs that are now closed, yay:
gnuradio/gnuradio#1928 (what vesrion is this reported against, master uses install prefix, possibly fixed by 78ba4b8794)
Notes on building:
CC="distcc gcc" CXX="distcc g++" cmake \
-DCMAKE_INSTALL_PREFIX=/tmp/gnuradio \
-DENABLE_DOXYGEN=OFF \
-DENABLE_NATIVE=ON \
-DENABLE_INTERNAL_VOLK=OFF \
.. && make install
CC="clang" CXX="clang++" cmake -DCMAKE_INSTALL_PREFIX=/tmp/gnuradio -DENABLE_DOXYGEN=OFF -DENABLE_NATIVE=ON -DPYTHON_EXECUTABLE=/usr/bin/python2 ..