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

[PATCH v2 0/6] Let the sequencer handle `git rebase -i --root`

From
Johannes Schindelin <johannes.schindelin@gmx.de>
Date
May 3, 2018, 23:01 UTC
Message-ID
<cover.1525388472.git.johannes.schindelin@gmx.de>
In-Reply-To
<cover.1524868165.git.johannes.schindelin@gmx.de>

When I reimplemented the most performance-critical bits of the interactive rebase in the sequencer, to speed up `git rebase -i` particularly on Windows (even if the benefits are still quite notable on Linux or macOS), I punted on the --root part.

I had always hoped that some other contributor (or I myself) would come back later to address the --root part in the sequencer, too, with the idea that this would move the last remaining complicated code from git-rebase--interactive.sh into sequencer.c, to facilitate converting the rest of git-rebase--interactive.sh.

When I say "the last remaining complicated code", of course I neglect the --preserve-merges code, but as I worked hard on the --rebase-merges patch series with the intention to eventually deprecate and maybe even remove the --preserve-merges mode, I always implicitly assume that the --preserve-merges code will be moved into its own shell script (git-rebase--preserve-merges.sh, maybe?) and never be converted.

So here goes: the patches to move the handling of --root into the sequencer. After two preparatory patches, the real conversion takes place in the third patch. After that, we take care of the --root related concerns that arise in conjunction with the --rebase-merges mode.

As the --rebase-merges/--root patches overlap quite a bit (not so much in the code itself as in philosophical considerations such as "what should happen if you try to merge a branch into a new root", or the fact that the label/reset/merge commands make it desirable to be able to create a new root commit in the middle of a todo list), I had to consider in which order to contribute them. In the end, I decided to go with --rebase-merges first, so the --root patches are based on the --rebase-merges patch series.

I consider this patch series a critical prerequisite for Alban's Google Summer of Code project to convert rebase -i into a builtin.

Changes since v1:
- Expanded is_pick_or_similar() to use a clear and verbose switch()
  statement, replacing the magic "command <= TODO_SQUASH".
- Completely revamped read_author_ident(), using sq_dequote(); Sadly,
  trying to reuse builtin/am.c's read_author_script() would result in
  substantialy more lines of code, and I also failed to find an easy way
  to allow for arbitrary order of the environment variables in
  author-script.
Johannes Schindelin (6):
  sequencer: extract helper to update active_cache_tree
  sequencer: learn about the special "fake root commit" handling
  rebase -i --root: let the sequencer handle even the initial part
  sequencer: allow introducing new root commits
  rebase --rebase-merges: a "merge" into a new root is a fast-forward
  rebase --rebase-merges: root commits can be cousins, too
 git-rebase--interactive.sh        |   4 +-
 sequencer.c                       | 206 ++++++++++++++++++++++++++----
 sequencer.h                       |   4 +
 t/t3404-rebase-interactive.sh     |  19 ++-
 t/t3421-rebase-topology-linear.sh |   6 +-
 t/t3430-rebase-merges.sh          |  72 +++++++++++
 6 files changed, 276 insertions(+), 35 deletions(-)
base-commit: 673fb9cb8b5c7d57cb560b6ade45e419c8dd09fc
Based-On: recreate-merges at https://github.com/dscho/git
Fetch-Base-Via: git fetch https://github.com/dscho/git recreate-merges
Published-As: https://github.com/dscho/git/releases/tag/sequencer-and-root-commits-v2
Fetch-It-Via: git fetch https://github.com/dscho/git sequencer-and-root-commits-v2
Branch-diff vs v1:
 1: 42db734a980 ! 1: 73398da7119 sequencer: learn about the special "fake root commit" handling
     @@ -54,40 +54,50 @@
       	return NULL;
       }
       
     ++/* Read author-script and return an ident line (author <email> timestamp) */
      +static const char *read_author_ident(struct strbuf *buf)
      +{
     -+	char *p, *p2;
     ++	const char *keys[] = {
     ++		"GIT_AUTHOR_NAME=", "GIT_AUTHOR_EMAIL=", "GIT_AUTHOR_DATE="
     ++	};
     ++	char *in, *out, *eol;
     ++	int i = 0, len;
      +
      +	if (strbuf_read_file(buf, rebase_path_author_script(), 256) <= 0)
      +		return NULL;
      +
     -+	for (p = buf->buf; *p; p++)
     -+		if (skip_prefix(p, "'\\\\''", (const char **)&p2))
     -+			strbuf_splice(buf, p - buf->buf, p2 - p, "'", 1);
     -+		else if (*p == '\'')
     -+			strbuf_splice(buf, p-- - buf->buf, 1, "", 0);
     -+
     -+	if (skip_prefix(buf->buf, "GIT_AUTHOR_NAME=", (const char **)&p)) {
     -+		strbuf_splice(buf, 0, p - buf->buf, "", 0);
     -+		p = strchr(buf->buf, '\n');
     -+		if (skip_prefix(p, "\nGIT_AUTHOR_EMAIL=", (const char **)&p2)) {
     -+			strbuf_splice(buf, p - buf->buf, p2 - p, " <", 2);
     -+			p = strchr(p, '\n');
     -+			if (skip_prefix(p, "\nGIT_AUTHOR_DATE=@",
     -+					(const char **)&p2)) {
     -+				strbuf_splice(buf, p - buf->buf, p2 - p,
     -+					      "> ", 2);
     -+				p = strchr(p, '\n');
     -+				if (p) {
     -+					strbuf_setlen(buf, p - buf->buf);
     -+					return buf->buf;
     -+				}
     -+			}
     ++	/* dequote values and construct ident line in-place */
     ++	for (in = out = buf->buf; i < 3 && in - buf->buf < buf->len; i++) {
     ++		if (!skip_prefix(in, keys[i], (const char **)&in)) {
     ++			warning("could not parse '%s' (looking for '%s'",
     ++				rebase_path_author_script(), keys[i]);
     ++			return NULL;
      +		}
     -+	}
     -+
     -+	warning(_("could not parse '%s'"), rebase_path_author_script());
     -+	return NULL;
     ++
     ++		eol = strchrnul(in, '\n');
     ++		*eol = '\0';
     ++		sq_dequote(in);
     ++		len = strlen(in);
     ++
     ++		if (i > 0) /* separate values by spaces */
     ++			*(out++) = ' ';
     ++		if (i == 1) /* email needs to be surrounded by <...> */
     ++			*(out++) = '<';
     ++		memmove(out, in, len);
     ++		out += len;
     ++		if (i == 1) /* email needs to be surrounded by <...> */
     ++			*(out++) = '>';
     ++		in = eol + 1;
     ++	}
     ++
     ++	if (i < 3) {
     ++		warning("could not parse '%s' (looking for '%s')",
     ++			rebase_path_author_script(), keys[i]);
     ++		return NULL;
     ++	}
     ++
     ++	buf->len = out - buf->buf;
     ++	return buf->buf;
      +}
      +
       static const char staged_changes_advice[] =
     @@ -159,7 +169,17 @@
      +/* Does this command create a (non-merge) commit? */
      +static int is_pick_or_similar(enum todo_command command)
      +{
     -+	return command <= TODO_SQUASH;
     ++	switch (command) {
     ++	case TODO_PICK:
     ++	case TODO_REVERT:
     ++	case TODO_EDIT:
     ++	case TODO_REWORD:
     ++	case TODO_FIXUP:
     ++	case TODO_SQUASH:
     ++		return 1;
     ++	default:
     ++		return 0;
     ++	}
      +}
      +
       static int update_squash_messages(enum todo_command command,
 2: 1c8740eaa91 = 2: 2dfe8315993 rebase -i --root: let the sequencer handle even the initial part
 3: 8a7f5751412 = 3: 0ee765bbb36 sequencer: allow introducing new root commits
 4: 0b7379b576b = 4: 29d7a9265e3 rebase --rebase-merges: a "merge" into a new root is a fast-forward
 5: 270b8fdf477 = 5: f7ee1b3de12 rebase --rebase-merges: root commits can be cousins, too
-- 
2.17.0.windows.1.38.g05ca542f78d
Previous: Johannes SchindelinNext: Johannes Schindelin
Message 14 of 22 in “Let the sequencer handle `git rebase -i --root`”
  1. 0/6 Let the sequencer handle `git rebase -i --root`Johannes Schindelin, Apr 27, 2018
  2. 1/6 sequencer: extract helper to update active_cache_treeJohannes Schindelin, Apr 27, 2018
  3. Stefan BellerApr 28, 2018
  4. 2/6 sequencer: learn about the special "fake root commit" handlingJohannes Schindelin, Apr 27, 2018
  5. Stefan BellerApr 28, 2018
  6. Johannes SchindelinApr 29, 2018
  7. Stefan BellerApr 29, 2018
  8. 3/6 rebase -i --root: let the sequencer handle even the initial partJohannes Schindelin, Apr 27, 2018
  9. Stefan BellerApr 28, 2018
  10. Johannes SchindelinApr 29, 2018
  11. 4/6 sequencer: allow introducing new root commitsJohannes Schindelin, Apr 27, 2018
  12. 5/6 rebase --rebase-merges: a "merge" into a new root is a fast-forwardJohannes Schindelin, Apr 27, 2018
  13. 6/6 rebase --rebase-merges: root commits can be cousins, tooJohannes Schindelin, Apr 27, 2018
  14. 0/6 Let the sequencer handle `git rebase -i --root`Johannes Schindelin, May 3, 2018
  15. 1/6 sequencer: extract helper to update active_cache_treeJohannes Schindelin, May 3, 2018
  16. 3/6 rebase -i --root: let the sequencer handle even the initial partJohannes Schindelin, May 3, 2018
  17. 2/6 sequencer: learn about the special "fake root commit" handlingJohannes Schindelin, May 3, 2018
  18. 5/6 rebase --rebase-merges: a "merge" into a new root is a fast-forwardJohannes Schindelin, May 3, 2018
  19. 6/6 rebase --rebase-merges: root commits can be cousins, tooJohannes Schindelin, May 3, 2018
  20. 4/6 sequencer: allow introducing new root commitsJohannes Schindelin, May 3, 2018
  21. Stefan BellerMay 4, 2018
  22. Johannes SchindelinMay 5, 2018

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.