[300/1136] Run test ‘/Users/louis/build/TextMate/Frameworks/scm/test_scm’… FAILED: /Users/louis/build/TextMate/Frameworks/scm/test_scm && touch /Users/louis/build/TextMate/Frameworks/scm/test_scm.run test_scm: 2 of 32 tests failed: /Users/louis/textmate/Frameworks/scm/tests/t_git.cc:88: Expected (wc.status("folder") == scm::status::unversioned), found (clean != untracked) /Users/louis/textmate/Frameworks/scm/tests/t_git.cc:187: Expected (wc.status("file") == scm::status::unversioned), found (clean != untracked) [300/1136] Compile ‘/Users/louis/build/TextMate/Frameworks/command/test_command.cc’… ninja: build stopped: subcommand failed.
After noting #283 https://github.com/textmate/textmate/issues/283, I deactivated my MacPorts git-core package. (I had already deactivated my MacPorts subversion package.)
After ninja -t clean && ./configure && ninja, the build completed successfully.
Hope this helps someone.
Louis
On 4 Jul 2013, at 13:51, Louis Zulli wrote:
[…] After noting #283 https://github.com/textmate/textmate/issues/283, I deactivated my MacPorts git-core package. (I had already deactivated my MacPorts subversion package.)
After ninja -t clean && ./configure && ninja, the build completed successfully.
This is a race condition bug of sorts. Basically all the SCM drivers rely on getting file system notifications for refreshing status, which introduces a random element — the test should handle it, but I have occasionally seen it fail myself, in which case re-running the build normally works.