Commit Briefs
Call pthread_cond_destroy(cond) exactly once when closing a view.
This moves the pthread_cond_destroy(need_commits) from stop_log_thread(), which can be called twice, to close_log_view(), which is called once. It also destroys the commit_loaded condition variable, which is created in open_log_view() but was never destroyed. ok stsp
merge new diff implementation from the git.gameoftrees.org diff.git repository
This new diff implementation was started by Neels Hofmeyr during the u2k20 hackathon and now replaces diffreg.c code lifted from the OpenBSD base system. The integration of this code into Got was done by me. Got now uses the patience diff algorithm by default. The diff.git repository will remain the primary repository for the diff code, which already compiles and runs on other operating systems such as Linux. Any fixes and improvements for files inherited from the diff.git repository should be written against that repository and synced to got.git afterwards.
changes for 0.44 (tags/0.44)
fix replacing symlinks with files
And add a test case which verifies that the inverse also works, i.e. a symlink being replaced with a regular file. problem reported and fix tested by jrick
avoid got_repo_map_path() in 'got tree' if a work tree is available
Note that got_worktree_resolve_path() will return a path based on the current working directory if the argument is the empty string. This quirk keeps the behaviour of 'got tree' intact which depends on the current working directory if run in a work tree. ok naddy@
test case for 'got integrate' failing if symlink changes to file; patch by jrick
This test verifies the current behaviour, even though it is not the most desirable behaviour.
avoid got_repo_map_path() in 'got blame'
'got blame' does not need access to the work tree. So far the work tree was completely hidden with unveil(). We must now expose the work tree while resolving the path, so unveil() calls are shuffled around slightly. Failing realpath() calls would mess with path resolution otherwise. There's a bug in got_worktree_resolve_path() where it failed to canonicalize a path constructed from a user-specified path that does not exist on disk. Note that this path falls into strncmp() a few lines down. I am fixing this by adding canonicalization. Generally, joining paths with asprintf() and comparing paths with strncmp() is fragile. A more general solution might be needed but I am leaving that for another day. ok naddy
preserve what was passed in if got_repo_map_path() can't figure out a path
problem found by + ok naddy