- fi
- test -z "$TEMPDIR" || rm -rf "$TEMPDIR"
- export PKG_CONFIG_PATH="$HOME/$libgit2dir/lib/pkgconfig:$PKG_CONFIG_PATH"
- export LD_LIBRARY_PATH="$HOME/$libgit2dir/lib:$LD_LIBRARY_PATH"
+before_script:
- # prepare for GP
- export CFLAGS="-g -O2 -Werror=pointer-arith -Werror=aggregate-return -Werror=implicit-function-declaration"
+script:
- NOCONFIGURE=1 ./autogen.sh &&
- mkdir _build &&
- cd _build &&
- ../configure &&
- make -j2 &&
Travis container-based build machines advertize 2 cores [in the docs](https://docs.travis-ci.com/user/migrating-from-legacy/#More-available-resour...), and legacy VM ones [1.5](https://docs.travis-ci.com/user/speeding-up-the-build/#Parallelizing-your-bu...). So actually for speed we could even go to 3 parallel jobs (or more) to account for time wasted in non-CPU-bound operations (file reads, etc.).
error output might be harder to read.
How so? Any sane Make implementation has output synchronization, and GNU Make's default is to group target output together, so I don't see the issue? But even if target-grouping is not enough in your opinion, we could choose recursive grouping (`--output-sync=recursive`).
--- Reply to this email directly or view it on GitHub: https://github.com/geany/geany-plugins/pull/394/files#r55345550