After #1151 and #1159 improvements to Makefile structure, some older systems (including some of those on Buildbot - across Linux, BSD and MacOS) no longer build the master branch, claiming issues with the out-of-dir sources, e.g. http://buildbot.networkupstools.org/public/nut/builders/Debian-x64-clang-stretch/builds/567/steps/compile/logs/stdio:
make[4]: Entering directory '/data/buildbot/slave/Debian-x64-clang-stretch/build/tools/nut-scanner'
Makefile:682: .deps/libnutscan_la-bcmxcp_ser.Plo: No such file or directory
Makefile:683: .deps/libnutscan_la-serial.Plo: No such file or directory
make[4]: *** No rule to make target '.deps/libnutscan_la-serial.Plo'. Stop.
make[4]: Leaving directory '/data/buildbot/slave/Debian-x64-clang-stretch/build/tools/nut-scanner'
Makefile:599: recipe for target 'distdir' failed
On main NUT CI farm, and on BuildBot worker with "buster", this does not happen.
Earlier there were issues with literal $(top_builddir) directory created during build, confusing the expected hierarchy. Here with "dist" being the problem, the "dist-hook" is worth investigating too.
After #1151 and #1159 improvements to Makefile structure, some older systems (including some of those on Buildbot - across Linux, BSD and MacOS) no longer build the master branch, claiming issues with the out-of-dir sources, e.g. http://buildbot.networkupstools.org/public/nut/builders/Debian-x64-clang-stretch/builds/567/steps/compile/logs/stdio:
On main NUT CI farm, and on BuildBot worker with "buster", this does not happen.
Earlier there were issues with literal
$(top_builddir)directory created during build, confusing the expected hierarchy. Here with "dist" being the problem, the "dist-hook" is worth investigating too.