git/dir.c, branch v2.34.2

Merge branch 'ds/add-rm-with-sparse-index' into maint

2021-11-23T22:48:11Z

Regression fix for 2.34 * ds/add-rm-with-sparse-index: dir: revert "dir: select directories correctly"

dir: revert "dir: select directories correctly"

2021-11-22T22:53:23Z

This reverts commit f6526728f950cacfd5b5e42bcc65f2c47f3da654. The change in f652672 (dir: select directories correctly, 2021-09-24) caused a regression in directory-based matches with non-cone-mode patterns, especially for .gitignore patterns. A test is included to prevent this regression in the future. The commit ed495847 (dir: fix pattern matching on dirs, 2021-09-24) was reverted in 5ceb663 (dir: fix directory-matching bug, 2021-11-02) for similar reasons. Neither commit changed tests, and tests added later in the series continue to pass when these commits are reverted. Reported-by: Danial Alihosseini Signed-off-by: Derrick Stolee Signed-off-by: Junio C Hamano

Merge branch 'ds/add-rm-with-sparse-index'

2021-11-03T20:32:28Z

Regression fix. * ds/add-rm-with-sparse-index: dir: fix directory-matching bug

dir: fix directory-matching bug

2021-11-03T17:10:36Z

This reverts the change from ed49584 (dir: fix pattern matching on dirs, 2021-09-24), which claimed to fix a directory-matching problem without a test case. It turns out to _create_ a bug, but it is a bit subtle. The bug would have been revealed by the first of two tests being added to t0008-ignores.sh. The first uses a pattern "/git/" inside the a/.gitignores file, which matches against 'a/git/foo' but not 'a/git-foo/bar'. This test would fail before the revert. The second test shows what happens if the test instead uses a pattern "git/" and this test passes both before and after the revert. The difference in these two cases are due to how last_matching_pattern_from_list() checks patterns both if they have the PATTERN_FLAG_MUSTBEDIR and PATTERN_FLAG_NODIR flags. In the case of "git/", the PATTERN_FLAG_NODIR is also provided, making the change in behavior in match_pathname() not affect the end result of last_matching_pattern_from_list(). Reported-by: Glen Choo Signed-off-by: Derrick Stolee Signed-off-by: Junio C Hamano

Merge branch 'mt/fix-add-rm-with-sparse-index'

2021-10-29T22:43:16Z

Fix-up to a topic already merged to 'master'. * mt/fix-add-rm-with-sparse-index: add, rm, mv: fix bug that prevents the update of non-sparse dirs

add, rm, mv: fix bug that prevents the update of non-sparse dirs

2021-10-28T15:46:07Z

These three commands recently learned to avoid updating paths outside the sparse checkout even if they are missing the SKIP_WORKTREE bit. This is done using path_in_sparse_checkout(), which checks whether a given path matches the current list of sparsity rules, similar to what clear_ce_flags() does when we run "git sparse checkout init" or "git sparse-checkout reapply". However, clear_ce_flags() uses a recursive approach, applying the match results from parent directories on paths that get the UNDECIDED result, whereas path_in_sparse_checkout() only attempts to match the full path and immediately considers UNDECIDED as NOT_MATCHED. This makes the function miss matches with leading directories. For example, if the user has the sparsity patterns "!/a" and "b/", add, rm, and mv will fail to update the path "a/b/c" and end up displaying a warning about it being outside the sparse checkout even though it isn't. This problem only occurs in full pattern mode as the pattern matching functions never return UNDECIDED for cone mode. To fix this, replicate the recursive behavior of clear_ce_flags() in path_in_sparse_checkout(), falling back to the parent directory match when a path gets the UNDECIDED result. (If this turns out to be too expensive in some cases, we may want to later add some form of caching to accelerate multiple queries within the same directory. This is not implemented in this patch, though.) Also add two tests for each affected command (add, rm, and mv) to check that they behave correctly with the recursive pattern matching. The first test would previously fail without this patch while the second already succeeded. It is added mostly to make sure that we are not breaking the existing pattern matching for directories that are really sparse, and also as a protection against any future regressions. Two other existing tests had to be changed as well: one test in t3602 checks that "git rm -r " won't remove sparse entries, but it didn't allow the non-sparse entries inside to be removed. The other one, in t7002, tested that "git mv" would correctly display a warning message for sparse paths, but it accidentally expected the message to include two non-sparse paths as well. Signed-off-by: Matheus Tavares Acked-by: Derrick Stolee Signed-off-by: Junio C Hamano

Merge branch 'ds/add-rm-with-sparse-index'

2021-10-13T22:15:56Z

"git add", "git mv", and "git rm" have been adjusted to avoid updating paths outside of the sparse-checkout definition unless the user specifies a "--sparse" option. * ds/add-rm-with-sparse-index: advice: update message to suggest '--sparse' mv: refuse to move sparse paths rm: skip sparse paths with missing SKIP_WORKTREE rm: add --sparse option add: update --renormalize to skip sparse paths add: update --chmod to skip sparse paths add: implement the --sparse option add: skip tracked paths outside sparse-checkout cone add: fail when adding an untracked sparse file dir: fix pattern matching on dirs dir: select directories correctly t1092: behavior for adding sparse files t3705: test that 'sparse_entry' is unstaged

dir: fix pattern matching on dirs

2021-09-28T17:31:02Z

Within match_pathname(), one successful matching category happens when the pattern is equal to its non-wildcard prefix. At this point, we have checked that the input 'pathname' matches the pattern up to the prefix length, and then we subtraced that length from both 'patternlen' and 'namelen'. In the case of a directory match, this prefix match should be sufficient. However, the success condition only cared about _exact_ equality here. Instead, we should allow any path that agrees on this prefix in the case of PATTERN_FLAG_MUSTBEDIR. This case was not tested before because of the way unpack_trees() would match a parent directory before visiting the contained paths. This approach is changing, so we must change this comparison. Signed-off-by: Derrick Stolee Signed-off-by: Junio C Hamano

dir: select directories correctly

2021-09-28T17:31:02Z

When matching a path against a list of patterns, the ones that require a directory match previously did not work when a filename is specified. This was fine when all pattern-matching was done within methods such as unpack_trees() that check a directory before recursing into the contained files. However, other commands will start matching individual files against pattern lists without that recursive approach. The last_matching_pattern_from_list() logic performs some checks on the filetype of a path within the index when the PATTERN_FLAG_MUSTBEDIR flag is set. This works great when setting SKIP_WORKTREE bits within unpack_trees(), but doesn't work well when passing an arbitrary path such as a file within a matching directory. We extract the logic around determining the file type, but attempt to avoid checking the filesystem if the parent directory already matches the sparse-checkout patterns. The new path_matches_dir_pattern() method includes a 'path_parent' parameter that is used to store the parent directory of 'pathname' between multiple pattern matching tests. This is loaded lazily, only on the first pattern it finds that has the PATTERN_FLAG_MUSTBEDIR flag. If we find that a path has a parent directory, we start by checking to see if that parent directory matches the pattern. If so, then we do not need to query the index for the type (which can be expensive). If we find that the parent does not match, then we still must check the type from the index for the given pathname. Note that this does not affect cone mode pattern matching, but instead the more general -- and slower -- full pattern set. Thus, this does not affect the sparse index. Helped-by: Ævar Arnfjörð Bjarmason Signed-off-by: Derrick Stolee Signed-off-by: Junio C Hamano

Merge branch 'jt/submodule-name-to-gitdir'

2021-09-23T20:44:49Z

Code refactoring. * jt/submodule-name-to-gitdir: submodule: extract path to submodule gitdir func