Commit Briefs

Stefan Sperling

use Patience diff for merging during cherrypick/backout/histedit/rebase

This has been shown to prevent mis-merges in some cases. It's probably not a final solution. We should look at what Git's "recursive merge" is doing and implement something similar. Keep using Myers during update/unstage. The advantage of Myers is that it produces smaller conflict chunks, and there are no known cases of mis-merges which affect update/unstage.


Stefan Sperling

check for close(2) error in got_repo_close() and propagate errors up

ok tracey


Stefan Sperling

fix unrelated changes being merged by got cherrypick/backout/rebase/histedit

This was a long-standing and very annoying bug. The two xfail tests in the cherrypick test suite are passing now.








Stefan Sperling

do not update symlinks which are already up-to-date

This fixes spurious 'U' notifications for symlinks during 'got update' that occurred even when the work tree was fully up-to-date. Observed on a work tree of the FreeBSD src repo and reproduced in our test suite by adding a no-op update at the end of a test which deals with updating symlinks.



Stefan Sperling

fix use of an uninitialized stat buffer during work tree status crawl

found by naddy ok naddy jrick


Stefan Sperling

ensure that old commits remain referenced after rebase and histedit

Create automatic "backup" references which ensure that objects from the pre-rebase or pre-histedit state remain in the repository. A new -l option for 'got rebase' and 'got histedit' lists old commits. This makes it easier to recover from botched rebase or histedit operations. Removal of such objects currently requires got ref -d and git-gc. This will be made more convenient in the future. testing and ok jrick


Stefan Sperling

implicitly mark all files in work tree as up-to-date after 'got integrate'

Avoids having to run 'got update' for no good reason after 'got integrate'. The same change was made recently for both rebase and histedit in commit a615e0e7796ea1103a6e0d4b5dbb6134597886660 and we forgot about histedit.


Stefan Sperling

close file handles before freeing other things in got_worktree_close()

The work tree's path needs to be valid while constructing error messages.



Stefan Sperling

make close(2) failure checks consistent; check 'close() == -1' everywhere

ok millert, naddy


Stefan Sperling

make fclose(3) failure checks consistent; check 'fclose() == EOF' everywhere

ok millert, naddy



Christian Weisgerber

fix a use after free()

ok jrick stsp


Christian Weisgerber

replace fparseln(3) with getline(3), for better portability

ok stsp


Stefan Sperling

implicitly mark all files in work tree as up-to-date after rebase/histedit

This should always be correct, since rebase and histedit start out with a clean and single-base-commit worktree, and end up committing all changes across the entire work tree when they are successful. tested by jrick and myself


Yang Zhong

add fd field to got_worktree, modify got_fileindex_entry_update to use fds

These changes are intended to make got more compatible with FreeBSD's Capsicum. ok stsp



Stefan Sperling

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.