Re: [PATCH v2] add-patch: edit the hunk again

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Rubén

Thanks for the re-roll. I'm still not convinced that changing this without keeping an easy way to get the current behavior is a good idea.

On 18/09/2024 18:51, Rubén Justo wrote:
The "edit" option allows the user to directly modify the hunk to be
applied.

If the modified hunk returned by the user is not an applicable patch,
they will be given the opportunity to try again.

For this new attempt we give them the original hunk;  they have to
repeat the modification from scratch.

Instead, let's give them the modified patch back, so they can identify
and fix the problem.

It's still not clear how an inexperienced user is meant to do that.

If they really want to start over with a fresh patch they still can
say "no" to cancel the "edit" and start anew [*].

This is not very obvious to the user, it would be much better to give them the choice when we prompt them about editing the hunk again. We've been giving the user the original hunk for the last six and a half years so I think it's a bit late to unilaterally change that now.

diff --git a/add-patch.c b/add-patch.c
index 557903310d..75b5129281 100644
--- a/add-patch.c
+++ b/add-patch.c
@@ -1111,7 +1111,8 @@ static void recolor_hunk(struct add_p_state *s, struct hunk *hunk)
  	hunk->colored_end = s->colored.len;
  }
-static int edit_hunk_manually(struct add_p_state *s, struct hunk *hunk)
+static int edit_hunk_manually(struct add_p_state *s, struct hunk *hunk,

I would add
				const struct hunk *backup,

here

+			      size_t plain_len, size_t colored_len)
  {
  	size_t i;
@@ -1146,6 +1147,10 @@ static int edit_hunk_manually(struct add_p_state *s, struct hunk *hunk)
  				      "addp-hunk-edit.diff", NULL) < 0)
  		return -1;
+ /* Drop possible previous edits */
+	strbuf_setlen(&s->plain, plain_len);
+	strbuf_setlen(&s->colored, colored_len);

then we can restore the back up here with

	*hunk = *backup;

That would make it clear that we're resetting the hunk and would continue to work if we change struct hunk in the future.

  	/* strip out commented lines */
  	hunk->start = s->plain.len;
  	for (i = 0; i < s->buf.len; ) {
@@ -1157,12 +1162,13 @@ static int edit_hunk_manually(struct add_p_state *s, struct hunk *hunk)
  	}
hunk->end = s->plain.len;
+
+	recolor_hunk(s, hunk);
+

This means we're now forking an external process when there is no hunk to color. It would be better to avoid that by leaving this code where it was and restoring the backup hunk above.

  	if (hunk->end == hunk->start)
  		/* The user aborted editing by deleting everything */
  		return 0;
- recolor_hunk(s, hunk);
-
>
  		/*
  		 * TRANSLATORS: do not translate [y/n]
@@ -1289,8 +1290,14 @@ static int edit_hunk_loop(struct add_p_state *s,
  					"Edit again (saying \"no\" discards!) "
  					"[y/n]? "));

I think we should make this a three-way choice so the user can choose to keep their changes or start from a valid hunk.

  		if (res < 1)
-			return -1;
+			break;
  	}
+
+	/* Drop a possible edit */
+	strbuf_setlen(&s->plain, plain_len);
+	strbuf_setlen(&s->colored, colored_len);
+	*hunk = backup;
+	return -1;
  }
static int apply_for_checkout(struct add_p_state *s, struct strbuf *diff,
diff --git a/t/t3701-add-interactive.sh b/t/t3701-add-interactive.sh
index 718438ffc7..f3206a317b 100755
--- a/t/t3701-add-interactive.sh
+++ b/t/t3701-add-interactive.sh
@@ -165,6 +165,19 @@ test_expect_success 'dummy edit works' '
  	diff_cmp expected diff
  '
+test_expect_success 'editing again works' '
+	git reset &&
+	write_script "fake_editor.sh" <<-\EOF &&
+	grep been-here "$1" >output
+	echo been-here >"$1"
+	EOF
+	(
+		test_set_editor "$(pwd)/fake_editor.sh" &&
+		test_write_lines e y | GIT_TRACE=1 git add -p

This is still missing "n q". Apart from that the test is looking good.

Best Wishes

Phillip

+	) &&
+	test_grep been-here output
+'
+
  test_expect_success 'setup patch' '
  	cat >patch <<-\EOF
  	@@ -1,1 +1,4 @@

Range-diff:
1:  bcf32d0979 ! 1:  2b55a759d5 add-patch: edit the hunk again
     @@ add-patch.c: static int edit_hunk_manually(struct add_p_state *s, struct hunk *h
       	/* strip out commented lines */
       	hunk->start = s->plain.len;
       	for (i = 0; i < s->buf.len; ) {
     +@@ add-patch.c: static int edit_hunk_manually(struct add_p_state *s, struct hunk *hunk)
     + 	}
     +
     + 	hunk->end = s->plain.len;
     ++
     ++	recolor_hunk(s, hunk);
     ++
     + 	if (hunk->end == hunk->start)
     + 		/* The user aborted editing by deleting everything */
     + 		return 0;
     +
     +-	recolor_hunk(s, hunk);
     +-
     + 	/*
     + 	 * If the hunk header is intact, parse it, otherwise simply use the
     + 	 * hunk header prior to editing (which will adjust `hunk->start` to
      @@ add-patch.c: static int edit_hunk_loop(struct add_p_state *s,
       	backup = *hunk;
@@ t/t3701-add-interactive.sh: test_expect_success 'dummy edit works' '
       	diff_cmp expected diff
       '
-+test_expect_success 'setup re-edit editor' '
     -+	write_script "fake_editor.sh" <<-\EOF &&
     -+	grep been-here "$1" && echo found >output
     -+	echo been-here > "$1"
     -+	EOF
     -+	test_set_editor "$(pwd)/fake_editor.sh"
     -+'
     -+
      +test_expect_success 'editing again works' '
      +	git reset &&
     -+	test_write_lines e y | GIT_TRACE=1 git add -p &&
     -+	grep found output
     ++	write_script "fake_editor.sh" <<-\EOF &&
     ++	grep been-here "$1" >output
     ++	echo been-here >"$1"
     ++	EOF
     ++	(
     ++		test_set_editor "$(pwd)/fake_editor.sh" &&
     ++		test_write_lines e y | GIT_TRACE=1 git add -p
     ++	) &&
     ++	test_grep been-here output
      +'
      +
       test_expect_success 'setup patch' '




[Index of Archives]     [Linux Kernel Development]     [Gcc Help]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [V4L]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]     [Fedora Users]

  Powered by Linux