git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* git status during interactive rebase
@ 2016-01-20 23:21 Stefan Beller
  2016-01-21  8:23 ` Matthieu Moy
  2016-01-21  9:28 ` Johannes Schindelin
  0 siblings, 2 replies; 6+ messages in thread
From: Stefan Beller @ 2016-01-20 23:21 UTC (permalink / raw)
  To: git@vger.kernel.org; +Cc: Guillaume Pages, Matthieu Moy

So I ran an interactive rebase, and while editing
.git/rebase-merge/git-rebase-todo I tried to run
`git status` in another terminal to inquire about a
filename of an untracked file.

However, I got:

$ git status
On branch submodule-groups-v3
fatal: Could not open file .git/rebase-merge/done for reading: No such
file or directory

Was this behavior always the case? (IIRC it worked a long time ago?)
Would it make sense to not abort completely but give a limited status?

Thanks,
Stefan

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: git status during interactive rebase
  2016-01-20 23:21 git status during interactive rebase Stefan Beller
@ 2016-01-21  8:23 ` Matthieu Moy
  2016-01-21 22:16   ` Stefan Beller
  2016-01-21  9:28 ` Johannes Schindelin
  1 sibling, 1 reply; 6+ messages in thread
From: Matthieu Moy @ 2016-01-21  8:23 UTC (permalink / raw)
  To: Stefan Beller; +Cc: git@vger.kernel.org, Guillaume Pages

Stefan Beller <sbeller@google.com> writes:

> So I ran an interactive rebase, and while editing
> .git/rebase-merge/git-rebase-todo I tried to run
> `git status` in another terminal to inquire about a
> filename of an untracked file.
>
> However, I got:
>
> $ git status
> On branch submodule-groups-v3
> fatal: Could not open file .git/rebase-merge/done for reading: No such
> file or directory
>
> Was this behavior always the case? (IIRC it worked a long time ago?)

>From the list of recipients, I guess you already found that the issue
probably comes from 84e6fb9 (status: give more information during rebase
-i, 2015-07-06), which introduced
read_rebase_todolist("rebase-merge/done", &have_done); in wt-status.c.

> Would it make sense to not abort completely but give a limited status?

Yes.

I guess read_rebase_todolist should be changed:

static void read_rebase_todolist(const char *fname, struct string_list *lines)
{
	struct strbuf line = STRBUF_INIT;
	FILE *f = fopen(git_path("%s", fname), "r");

	if (!f)
		die_errno("Could not open file %s for reading",
			  git_path("%s", fname));

This should return an empty string_list instead of dying when the file
does not exist. The case of an empty list is already dealt with later:

		if (have_done.nr == 0)
			status_printf_ln(s, color, _("No commands done."));

No time to work on a patch for now :-(.

-- 
Matthieu Moy
http://www-verimag.imag.fr/~moy/

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: git status during interactive rebase
  2016-01-20 23:21 git status during interactive rebase Stefan Beller
  2016-01-21  8:23 ` Matthieu Moy
@ 2016-01-21  9:28 ` Johannes Schindelin
  1 sibling, 0 replies; 6+ messages in thread
From: Johannes Schindelin @ 2016-01-21  9:28 UTC (permalink / raw)
  To: Stefan Beller; +Cc: git@vger.kernel.org, Guillaume Pages, Matthieu Moy

Hi Stefan,

On Wed, 20 Jan 2016, Stefan Beller wrote:

> So I ran an interactive rebase, and while editing
> .git/rebase-merge/git-rebase-todo I tried to run
> `git status` in another terminal to inquire about a
> filename of an untracked file.

Heh, I don't think that anybody did that before, because the rebase has
not even quite started yet...

The cop-out would be to write an empty 'done' file before editing the
todo, but it would give the wrong impression that it is safe to run `git
rebase --continue` now... Interactive rebase is *definitely* not
thread-safe ;-)

So the proper fix might be to test for the presence of the "done" file and
otherwise tell the user that this rebase has not even started yet.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: git status during interactive rebase
  2016-01-21  8:23 ` Matthieu Moy
@ 2016-01-21 22:16   ` Stefan Beller
  2016-01-21 22:35     ` Junio C Hamano
  0 siblings, 1 reply; 6+ messages in thread
From: Stefan Beller @ 2016-01-21 22:16 UTC (permalink / raw)
  To: Matthieu Moy; +Cc: git@vger.kernel.org, Guillaume Pages

On Thu, Jan 21, 2016 at 12:23 AM, Matthieu Moy
<Matthieu.Moy@grenoble-inp.fr> wrote:
> Stefan Beller <sbeller@google.com> writes:
>
>> So I ran an interactive rebase, and while editing
>> .git/rebase-merge/git-rebase-todo I tried to run
>> `git status` in another terminal to inquire about a
>> filename of an untracked file.
>>
>> However, I got:
>>
>> $ git status
>> On branch submodule-groups-v3
>> fatal: Could not open file .git/rebase-merge/done for reading: No such
>> file or directory
>>
>> Was this behavior always the case? (IIRC it worked a long time ago?)
>
> From the list of recipients, I guess you already found that the issue
> probably comes from 84e6fb9 (status: give more information during rebase
> -i, 2015-07-06), which introduced
> read_rebase_todolist("rebase-merge/done", &have_done); in wt-status.c.

I just remembered you working on git-status specially during rebase,
so I figured you may have a clue here.

> No time to work on a patch for now :-(.

:-(

> The cop-out would be to write an empty 'done' file before editing the
> todo, but it would give the wrong impression that it is safe to run `git
> rebase --continue` now... Interactive rebase is *definitely* not
> thread-safe ;-)

ok. As I did not start the rebase yet, I figured I may still ask a thing. :P

> So the proper fix might be to test for the presence of the "done" file and
> otherwise tell the user that this rebase has not even started yet.

So what Matthieu said?

Thanks,
Stefan

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: git status during interactive rebase
  2016-01-21 22:16   ` Stefan Beller
@ 2016-01-21 22:35     ` Junio C Hamano
  2016-01-22 16:28       ` Johannes Schindelin
  0 siblings, 1 reply; 6+ messages in thread
From: Junio C Hamano @ 2016-01-21 22:35 UTC (permalink / raw)
  To: Stefan Beller; +Cc: Matthieu Moy, git@vger.kernel.org, Guillaume Pages

Stefan Beller <sbeller@google.com> writes:

>> So the proper fix might be to test for the presence of the "done" file and
>> otherwise tell the user that this rebase has not even started yet.
>
> So what Matthieu said?

Yup, I think that is the right thing to do.

Perhaps something along this line (not tested and done on 'pu', so I
am not committing it to anywhere myself)?

 wt-status.c | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/wt-status.c b/wt-status.c
index ab4f80d..90b2474 100644
--- a/wt-status.c
+++ b/wt-status.c
@@ -1068,14 +1068,17 @@ static void abbrev_sha1_in_line(struct strbuf *line)
 
 }
 
-static void read_rebase_todolist(const char *fname, struct string_list *lines)
+static int read_rebase_todolist(const char *fname, struct string_list *lines)
 {
 	struct strbuf line = STRBUF_INIT;
 	FILE *f = fopen(git_path("%s", fname), "r");
 
-	if (!f)
+	if (!f) {
+		if (errno == ENOENT)
+			return -1;
 		die_errno("Could not open file %s for reading",
 			  git_path("%s", fname));
+	}
 	while (!strbuf_getline_lf(&line, f)) {
 		if (line.len && line.buf[0] == comment_line_char)
 			continue;
@@ -1085,6 +1088,7 @@ static void read_rebase_todolist(const char *fname, struct string_list *lines)
 		abbrev_sha1_in_line(&line);
 		string_list_append(lines, line.buf);
 	}
+	return 0;
 }
 
 static void show_rebase_information(struct wt_status *s,
@@ -1098,12 +1102,12 @@ static void show_rebase_information(struct wt_status *s,
 		struct string_list have_done = STRING_LIST_INIT_DUP;
 		struct string_list yet_to_do = STRING_LIST_INIT_DUP;
 
-		read_rebase_todolist("rebase-merge/done", &have_done);
-		read_rebase_todolist("rebase-merge/git-rebase-todo", &yet_to_do);
-
-		if (have_done.nr == 0)
+		if ((read_rebase_todolist("rebase-merge/done", &have_done) < 0) ||
+		    (read_rebase_todolist("rebase-merge/git-rebase-todo", &yet_to_do) < 0)) {
+			status_printf_ln(s, color, _("rebase-i not started yet."));
+		} else if (have_done.nr == 0) {
 			status_printf_ln(s, color, _("No commands done."));
-		else {
+		} else {
 			status_printf_ln(s, color,
 				Q_("Last command done (%d command done):",
 					"Last commands done (%d commands done):",

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: git status during interactive rebase
  2016-01-21 22:35     ` Junio C Hamano
@ 2016-01-22 16:28       ` Johannes Schindelin
  0 siblings, 0 replies; 6+ messages in thread
From: Johannes Schindelin @ 2016-01-22 16:28 UTC (permalink / raw)
  To: Junio C Hamano
  Cc: Stefan Beller, Matthieu Moy, git@vger.kernel.org, Guillaume Pages

Hi,

[Stefan, please do not drop me from the Cc: list ;-)]

On Thu, 21 Jan 2016, Junio C Hamano wrote:

> Perhaps something along this line (not tested and done on 'pu', so I
> am not committing it to anywhere myself)?

I like it!

I tested it, amended it and am about to send out a mail with a commit
message.

Ciao,
Dscho

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2016-01-22 16:28 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-01-20 23:21 git status during interactive rebase Stefan Beller
2016-01-21  8:23 ` Matthieu Moy
2016-01-21 22:16   ` Stefan Beller
2016-01-21 22:35     ` Junio C Hamano
2016-01-22 16:28       ` Johannes Schindelin
2016-01-21  9:28 ` Johannes Schindelin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).