git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[RFC/PATCH] fast-import: treat filemodify with empty tree as delete

From
Jonathan Nieder <jrnieder@gmail.com>
Date
Dec 11, 2010, 22:47 UTC
Message-ID
<20101211224732.GA18822@burratino>
In-Reply-To
<20101211184654.GA17464@burratino>
Jonathan Nieder wrote:
> Maybe fast-import should make the check itself, to avoid
> writing trees with empty directories that would be hard to
> re-create with "git write-tree".
Like this, maybe?
-- 8< --
Subject: fast-import: treat filemodify with empty tree as delete

Traditionally, git trees do not contain entries for empty subdirectories. Generally speaking, subtrees are not created or destroyed explicitly; instead, they automatically appear when needed to hold regular files, symlinks, and submodules.

v1.7.3-rc0~75^2 (Teach fast-import to import subtrees named by tree id, 2010-06-30) changed that, by allowing an empty subtree to be included in a fast-import stream explicitly:

	M 040000 4b825dc642cb6eb9a060e54bf8d69288fbee4904 subdir

That was unintentional. Better and more closely analogous to "git read-tree --prefix" to treat such an input line as a request to delete ("to empty") subdir.

Noticed-by: David Barr <david.barr@cordelta.com>
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
---
Tests use the "ls" command from vcs-svn-pu.  The actual change
would apply cleanly to master or maint, though.
 fast-import.c          |   10 ++++
 t/t9300-fast-import.sh |  107 ++++++++++++++++++++++++++++++++++++++++++++----
 2 files changed, 109 insertions(+), 8 deletions(-)
diff --git a/fast-import.c b/fast-import.c
index e62f34d..c774893 100644
--- a/fast-import.c
+++ b/fast-import.c
@@ -2196,6 +2196,16 @@ static void file_change_m(struct branch *b)
 		p = uq.buf;
 	}
 
+	/*
+	 * Git does not track empty, non-toplevel directories.
+	 */
+	if (S_ISDIR(mode) &&
+	    !memcmp(sha1, (unsigned char *) EMPTY_TREE_SHA1_BIN, 20) &&
+	    *p) {
+		tree_content_remove(&b->branch_tree, p, NULL);
+		return;
+	}
+
 	if (S_ISGITLINK(mode)) {
 		if (inline_data)
 			die("Git links cannot be specified 'inline': %s",
diff --git a/t/t9300-fast-import.sh b/t/t9300-fast-import.sh
index b0e3bda..c17f704 100755
--- a/t/t9300-fast-import.sh
+++ b/t/t9300-fast-import.sh
@@ -25,6 +25,14 @@ echo "$@"'
 
 >empty
 
+test_expect_success 'setup: have pipes?' '
+	rm -f frob &&
+	if mkfifo frob
+	then
+		test_set_prereq PIPE
+	fi
+'
+
 ###
 ### series A
 ###
@@ -881,6 +889,97 @@ test_expect_success \
 	 git diff-tree -C --find-copies-harder -r N4^ N4 >actual &&
 	 compare_diff_raw expect actual'
 
+test_expect_success PIPE 'N: read and copy directory' '
+	cat >expect <<-\EOF
+	:100755 100755 f1fb5da718392694d0076d677d6d0e364c79b0bc f1fb5da718392694d0076d677d6d0e364c79b0bc C100	file2/newf	file3/newf
+	:100644 100644 7123f7f44e39be127c5eb701e5968176ee9d78b1 7123f7f44e39be127c5eb701e5968176ee9d78b1 C100	file2/oldf	file3/oldf
+	EOF
+	git update-ref -d refs/heads/N4 &&
+	rm -f backflow &&
+	mkfifo backflow &&
+	(
+		exec <backflow &&
+		cat <<-EOF &&
+		commit refs/heads/N4
+		committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
+		data <<COMMIT
+		copy by tree hash, part 2
+		COMMIT
+
+		from refs/heads/branch^0
+		ls "file2"
+		EOF
+		read mode type tree filename &&
+		echo "M 040000 $tree file3"
+	) |
+	git fast-import --cat-blob-fd=3 3>backflow &&
+	git diff-tree -C --find-copies-harder -r N4^ N4 >actual &&
+	compare_diff_raw expect actual
+'
+
+test_expect_success PIPE 'N: read and copy "empty" directory' '
+	cat <<-\EOF >expect &&
+	OBJNAME
+	:000000 100644 OBJNAME OBJNAME A	greeting
+	OBJNAME
+	:100644 000000 OBJNAME OBJNAME D	unrelated
+	OBJNAME
+	:000000 100644 OBJNAME OBJNAME A	unrelated
+	EOF
+	git update-ref -d refs/heads/copy-empty &&
+	rm -f backflow &&
+	mkfifo backflow &&
+	(
+		exec <backflow &&
+		cat <<-EOF &&
+		commit refs/heads/copy-empty
+		committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
+		data <<COMMIT
+		copy "empty" (missing) directory
+		COMMIT
+
+		M 100644 inline src/greeting
+		data <<BLOB
+		hello
+		BLOB
+		C src/greeting dst1/non-greeting
+		C src/greeting unrelated
+		# leave behind "empty" src directory
+		D src/greeting
+		ls "src"
+		EOF
+		read mode type tree filename &&
+		sed -e "s/X\$//" <<-EOF
+		M $mode $tree dst1
+		M $mode $tree dst2
+
+		commit refs/heads/copy-empty
+		committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
+		data <<COMMIT
+		copy empty directory to root
+		COMMIT
+
+		M $mode $tree X
+
+		commit refs/heads/copy-empty
+		committer $GIT_COMMITTER_NAME <$GIT_COMMITTER_EMAIL> $GIT_COMMITTER_DATE
+		data <<COMMIT
+		add another file
+		COMMIT
+
+		M 100644 inline greeting
+		data <<BLOB
+		hello
+		BLOB
+		EOF
+	) |
+	git fast-import --cat-blob-fd=3 3>backflow &&
+	git rev-list copy-empty |
+	git diff-tree -r --root --stdin |
+	sed "s/$_x40/OBJNAME/g" >actual &&
+	test_cmp expect actual
+'
+
 test_expect_success \
 	'N: copy root directory by tree hash' \
 	'cat >expect <<-\EOF &&
@@ -1773,14 +1872,6 @@ test_expect_success 'R: print two blobs to stdout' '
 	test_cmp expect actual
 '
 
-test_expect_success 'setup: have pipes?' '
-	rm -f frob &&
-	if mkfifo frob
-	then
-		test_set_prereq PIPE
-	fi
-'
-
 test_expect_success PIPE 'R: copy using cat-file' '
 	expect_id=$(git hash-object big) &&
 	expect_len=$(wc -c <big) &&

-- 
1.7.2.4
Previous: Jonathan NiederNext: Jonathan Nieder
Message 36 of 37 in “vcs-svn: prepare for (implement?) incremental import”
  1. 00/10 vcs-svn: prepare for (implement?) incremental importJonathan Nieder, Dec 10, 2010
  2. 01/10 vcs-svn: use higher mark numbers for blobsJonathan Nieder, Dec 10, 2010
  3. 02/10 vcs-svn: save marks for imported commitsJonathan Nieder, Dec 10, 2010
  4. Jonathan NiederMar 6, 2011
  5. 03/10 vcs-svn: introduce cat_mark function to retrieve a marked blobJonathan Nieder, Dec 10, 2010
  6. 04/10 vcs-svn: make apply_delta caller retrieve preimageJonathan Nieder, Dec 10, 2010
  7. 05/10 vcs-svn: split off function to export result from delta applicationJonathan Nieder, Dec 10, 2010
  8. 06/10 vcs-svn: do not rely on marks for old blobsJonathan Nieder, Dec 10, 2010
  9. 07/10 vcs-svn: split off function to make 'ls' requestsJonathan Nieder, Dec 10, 2010
  10. 08/10 vcs-svn: prepare to eliminate repo_tree structureJonathan Nieder, Dec 10, 2010
  11. vcs-svn: prepare to eliminate repo_tree structureJonathan Nieder, Mar 6, 2011
  12. David BarrMar 6, 2011
  13. 09/10 vcs-svn: simplifications for repo_modify_path et alJonathan Nieder, Dec 10, 2010
  14. 10/10 vcs-svn: eliminate repo_tree structureJonathan Nieder, Dec 10, 2010
  15. 12/10 vcs-svn: quote paths correctly for ls commandDavid Michael Barr, Dec 11, 2010
  16. vcs-svn: quote all paths passed to fast-importJonathan Nieder, Dec 11, 2010
  17. 13/10 vcs-svn: use mark from previous import for parent commitDavid Michael Barr, Dec 12, 2010
  18. Jonathan NiederDec 12, 2010
  19. 00/12 vcs-svn: incremental importJonathan Nieder, Mar 6, 2011
  20. 01/12 vcs-svn: use higher mark numbers for blobsJonathan Nieder, Mar 6, 2011
  21. Junio C HamanoMar 8, 2011
  22. Jonathan NiederMar 9, 2011
  23. 02/12 vcs-svn: save marks for imported commitsJonathan Nieder, Mar 6, 2011
  24. 03/12 vcs-svn: introduce repo_read_path to check the content at a pathJonathan Nieder, Mar 6, 2011
  25. 04/12 vcs-svn: handle_node: use repo_read_pathJonathan Nieder, Mar 6, 2011
  26. 05/12 vcs-svn: simplify repo_modify_path and repo_copyJonathan Nieder, Mar 6, 2011
  27. 06/12 vcs-svn: add a comment before each commitJonathan Nieder, Mar 6, 2011
  28. 07/12 vcs-svn: allow input errors to be detected promptlyJonathan Nieder, Mar 6, 2011
  29. 08/12 vcs-svn: set up channel to read fast-import cat-blob responseJonathan Nieder, Mar 6, 2011
  30. 09/12 vcs-svn: eliminate repo_tree structureJonathan Nieder, Mar 6, 2011
  31. 10/12 vcs-svn: quote paths correctly for ls commandJonathan Nieder, Mar 6, 2011
  32. 11/12 vcs-svn: handle filenames with dq correctlyJonathan Nieder, Mar 6, 2011
  33. 12/12 vcs-svn: use mark from previous import for parent commitJonathan Nieder, Mar 6, 2011
  34. Sverre RabbelierMar 7, 2011
  35. Jonathan NiederMar 7, 2011
  36. fast-import: treat filemodify with empty tree as deleteJonathan Nieder, Dec 11, 2010
  37. vcs-svn: avoid git-isms in fast-import streamJonathan Nieder, Dec 11, 2010

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.