On 2019/11/07 14:36, Tracey Emery wrote: > > > +pre-build: > > > + @sed -i "s,-lrt,," \ > > > + ${WRKDIR}/build-amd64/build.ninja > > New patches to address this issue. I figured out where to make the > changes.
Thanks. > > > > this change is bogus, the input files must be fixed instead. > > > > > + ${MODPY_BIN} ${MODPY_LIBDIR}/compileall.py \ > > > > you can give this a directory, it will recurse > > Cool! I did not know this. That's nifty. > > Ok, so I made the changes and did another build. I did not experience > any errors like you mentioned, so I don't know. I've attached the ouput > of make fake. Same error again here. Other people had problems with parallel builds on Linux (I found some 2014-ish list posts) with the same thing. It looks like there's some scripting in the common/CMakeLists.txt that seems intended to make this work but it doesn't do the trick. ----------- # auto-generate netlist_lexer.h and netlist_keywords.cpp make_lexer( netlist_lexer_source_files ${CMAKE_CURRENT_SOURCE_DIR}/netlist.keywords ${PROJECT_SOURCE_DIR}/include/netlist_lexer.h ${CMAKE_CURRENT_SOURCE_DIR}/netlist_keywords.cpp NL_T # Pass header file with dependency on *_lexer.h as extra_arg ${CMAKE_PROJECT_SOURCE_DIR}/pcbnew/netlist_reader.h ) add_dependencies( common netlist_lexer_source_files ) add_dependencies( pcbcommon netlist_lexer_source_files ) ----------- However with pre-build removed we can use cmake's make backend which is less susceptible to these problems so I've done that for now and committed. > Also, attached is a new tarball with the patches and Makefile changes. > Below is also a diff. I didn't know what you wanted from this point > forward. It didn't include the added files, but I've regenerated from the tar.gz so don't worry about that.