patch, 2 partsFix a couple of minor rebase -i regressions

From: Phillip Wood <phillip.wood@dunelm.org.uk>

These patches fix a regression in the reflog message written when rebase -i finishes and supresses the status output from applying any autostashed changes to match the shell version of rebase -i.

Phillip Wood (2):
  rebase -i: fix reflog message
  rebase -i: silence stash apply

 sequencer.c | 3 +++
 1 file changed, 3 insertions(+)

From: Phillip Wood <phillip.wood@dunelm.org.uk>

When rebase -i was converted to C a bug was introduced into the code that creates the reflog message. Instead of saying rebase -i (finish): <head-name> onto <onto> it says rebase -i (finish): <head-name> onto <orig-head><onto> as the strbuf is not reset between reading the value of <orig-head> and <onto>.

Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
---
 sequencer.c | 1 +
 1 file changed, 1 insertion(+)

From: Phillip Wood <phillip.wood@dunelm.org.uk>

The shell version of rebase -i silences the status output from 'git stash apply' when restoring the autostashed changes. The C version does not.

Having the output from git stash apply on the screen is distracting as it makes it difficult to find the message from git rebase saying that the rebase succeeded. Also the status information that git stash prints talks about looking in .git/rebase-merge/done to see which commits have been applied. As .git/rebase-merge is removed shortly after the message is printed before rebase -i exits this is confusing.

Signed-off-by: Phillip Wood <phillip.wood@dunelm.org.uk>
---
 sequencer.c | 2 ++
 1 file changed, 2 insertions(+)

Hi Phillip,

On Thu, 18 May 2017, Phillip Wood wrote:

If I remember correctly, then the shell version prints the output in case of an error.

We already imitated that behavior in `git commit` (https://github.com/git-for-windows/git/blob/v2.13.0.windows.1/sequencer.c#L674-L684):

		/* hide stderr on success */
		struct strbuf buf = STRBUF_INIT;
		int rc = pipe_command(&cmd,
				      NULL, 0,
				      /* stdout is already redirected */
				      NULL, 0,
				      &buf, 0);
		if (rc)
			fputs(buf.buf, stderr);
		strbuf_release(&buf);

I think that would be the appropriate approach here, too.

Ciao, Johannes

P.S.: it may be a very good idea to accompany this patch (as well as the previous one) by a patch to the test suite to verify that the fixed code does not regress.

Hi Johannes

Thanks for your reply On 18/05/17 11:49, Johannes Schindelin wrote:

The shell version prints it's own error message if there's an error, the C version does this as well

Shell version:
apply_autostash () {
	if test -f "$state_dir/autostash"
	then
		stash_sha1=$(cat "$state_dir/autostash")
		if git stash apply $stash_sha1 2>&1 >/dev/null
		then
			echo "$(gettext 'Applied autostash.')"
		else
			git stash store -m "autostash" -q $stash_sha1 ||
			die "$(eval_gettext "Cannot store \$stash_sha1")"
			gettext 'Applying autostash resulted in conflicts.
Your changes are safe in the stash.
You can run "git stash pop" or "git stash drop" at any time.
'
		fi
	fi
}

C version:
static int apply_autostash(struct replay_opts *opts)
{
	struct strbuf stash_sha1 = STRBUF_INIT;
	struct child_process child = CHILD_PROCESS_INIT;
	int ret = 0;

	if (!read_oneliner(&stash_sha1, rebase_path_autostash(), 1)) {
		strbuf_release(&stash_sha1);
		return 0;
	}
	strbuf_trim(&stash_sha1);

	child.git_cmd = 1;
	child.no_stdout = 1;
	child.no_stderr = 1;
	argv_array_push(&child.args, "stash");
	argv_array_push(&child.args, "apply");
	argv_array_push(&child.args, stash_sha1.buf);
	if (!run_command(&child))
		printf(_("Applied autostash.\n"));
	else {
		struct child_process store = CHILD_PROCESS_INIT;

		store.git_cmd = 1;
		argv_array_push(&store.args, "stash");
		argv_array_push(&store.args, "store");
		argv_array_push(&store.args, "-m");
		argv_array_push(&store.args, "autostash");
		argv_array_push(&store.args, "-q");
		argv_array_push(&store.args, stash_sha1.buf);
		if (run_command(&store))
			ret = error(_("cannot store %s"), stash_sha1.buf);
		else
			printf(_("Applying autostash resulted in conflicts.\n"
				"Your changes are safe in the stash.\n"
				"You can run \"git stash pop\" or"
				" \"git stash drop\" at any time.\n"));
	}

	strbuf_release(&stash_sha1);
	return ret;
}

I can change it to do this if you think it is an improvement but the existing error message seems fairly clear to me and I think (though I haven't checked) that the call to stash store should print out the id of the stash for the user

I agree that a test for this (and probably my other changes) would be useful, I'll try and put something together but it won't be until after the end of next week.

Thanks

Phillip

Hi Phillip,

On Thu, 18 May 2017, Phillip Wood wrote:

Oh, you're right, I misremembered.

Thank you for being so thorough. This patch is Acked-by: me, then.

Thanks, Johannes