messages confusing
Hi all,
I had those messages today
% git merge origin/master
Already uptodate!
Merge made by recursive.
%
with git version 1.7.2.rc1.210.g7b476 and I wonder if it is really intended. It did create an empty merge commit but I find both messages together a bit confusing here. Why would Git merge if uptodate?
I don't think I will have time to investigate in the comming weeks but you should be able to reproduce it by following these steps (it is by now, at least):
% git clone git://github.com/nvie/gitflow.git
% cd gitflow
% git merge origin/master
%
Thanks,
Signature · 1 line
Nicolas Sebrecht
Quoted 6 lines · Nicolas Sebrecht
I had those messages today
% git merge origin/master
Already uptodate!
Merge made by recursive.
%
This is quite an old message; I don't think it has much to do with 1.7-ness of your version.
The first "uptodate" refers to your tree state being up-to-date. You already had the necessary changes the other history wanted you to have---they were acquired in your history before you started this merge by some other means. Perhaps merging from other trees, or picking up equivalent patches from mailing list, etc.
A merge commit is still created in this case, in order to record the fact that everything origin/master wanted to do to your history has already been reconciled to your history. Otherwise "git log ..origin/master" will still show commits your history does not have.
Quoted 9 lines · Junio C Hamano
This is quite an old message; I don't think it has much to do with 1.7-ness of your version.
The first "uptodate" refers to your tree state being up-to-date. You already had the necessary changes the other history wanted you to have---they were acquired in your history before you started this merge by some other means. Perhaps merging from other trees, or picking up equivalent patches from mailing list, etc.
This is what I was guessing but wanted to be sure that the resulting messages were intended as is.
Quoted 4 lines
A merge commit is still created in this case, in order to record the fact that everything origin/master wanted to do to your history has already been reconciled to your history. Otherwise "git log ..origin/master" will still show commits your history does not have.
Thanks for the explanations.
Signature · 1 line
Nicolas Sebrecht
Quoted 4 lines · Nicolas Sebrecht
Hi all,
I had those messages today
Quoted 8 lines
% git merge origin/master
Already uptodate!
Merge made by recursive.
%with git version 1.7.2.rc1.210.g7b476 and I wonder if it is really intended. It did create an empty merge commit but I find both messages together a bit confusing here. Why would Git merge if uptodate?
In any case, "uptodate" should be written as "up-to-date" for consistency with other user-visible messages in Git and in the docs.
Wincent
PATCH 1/1 use "up-to-date" in messages instead of "uptodate" for consistency
--- merge-recursive.c | 2 +- unpack-trees.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-)
merge-recursive.c +1 -1
@@ -1214,7 +1214,7 @@ int merge_trees(struct merge_options *o, }if (sha_eq (common->object.sha1, merge->object.sha1)) { -output (o,0 ,"Already uptodate!" ); +output (o,0 ,"Already up-to-date!" ); *result = head;return 1 ; }
unpack-trees.c +2 -2
@@ -22,7 +22,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' would be overwritten by merge. Cannot merge." ,/* not_uptodate_file */ -"Entry '%s' not uptodate. Cannot merge." , +"Entry '%s' not up-to-date. Cannot merge." ,/* not_uptodate_dir */ "Updating '%s' would lose untracked files in it",
@@ -34,7 +34,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' overlaps with '%s'. Cannot bind." ,/* sparse_not_uptodate_file */ -"Entry '%s' not uptodate. Cannot update sparse checkout." , +"Entry '%s' not up-to-date. Cannot update sparse checkout." ,/* would_lose_orphaned */ "Working tree file '%s' would be %s by sparse checkout update.",
Signature · 1 line
1.7.2.rc1.210.g7b476
Re: [PATCH] use "up-to-date" in messages instead of "uptodate" for consistency
Did you run the tests after applying this patch? This looks like it would break:
t/t7110-reset-merge.sh
176: grep file1 err.log | grep "not uptodate" 192: grep file1 err.log | grep "not uptodate"
PATCH v2 1/1 use "up-to-date" instead of "uptodate" for consistency
---
Quoted 5 lines · Ævar Arnfjörð Bjarmason
Did you run the tests after applying this patch? This looks like it would break:
t/t7110-reset-merge.sh176: grep file1 err.log | grep "not uptodate" 192: grep file1 err.log | grep "not uptodate"
I forgot about them. Thanks for a reminder.
Documentation/git-checkout.txt | 2 +- merge-recursive.c | 2 +- t/t7110-reset-merge.sh | 4 ++-- unpack-trees.c | 4 ++-- 4 files changed, 6 insertions(+), 6 deletions(-)
Documentation/git-checkout.txt +1 -1
@@ -263,7 +263,7 @@ the above checkout would fail like this: + ------------ $ git checkout mytopic -fatal: Entry 'frotz' not uptodate. Cannot merge. +fatal: Entry 'frotz' not up-to-date. Cannot merge. ------------ + You can give the `-m` flag to the command, which would try a
merge-recursive.c +1 -1
@@ -1214,7 +1214,7 @@ int merge_trees(struct merge_options *o, }if (sha_eq (common->object.sha1, merge->object.sha1)) { -output (o,0 ,"Already uptodate!" ); +output (o,0 ,"Already up-to-date!" ); *result = head;return 1 ; }
t/t7110-reset-merge.sh +2 -2
@@ -173,7 +173,7 @@ test_expect_success 'reset --merge fails with changes in file it touches' ' sed -e "s/line 1/changed line 1/" <file1 >file3 && mv file3 file1 && test_must_fail git reset --merge HEAD^ 2>err.log && - grep file1 err.log | grep "not uptodate" + grep file1 err.log | grep "not up-to-date" ' # The next test will test the following:
@@ -189,7 +189,7 @@ test_expect_success 'reset --keep fails with changes in file it touches' ' sed -e "s/line 1/changed line 1/" <file1 >file3 && mv file3 file1 && test_must_fail git reset --keep HEAD^ 2>err.log && - grep file1 err.log | grep "not uptodate" + grep file1 err.log | grep "not up-to-date" ' test_expect_success 'setup 3 different branches' '
unpack-trees.c +2 -2
@@ -22,7 +22,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' would be overwritten by merge. Cannot merge." ,/* not_uptodate_file */ -"Entry '%s' not uptodate. Cannot merge." , +"Entry '%s' not up-to-date. Cannot merge." ,/* not_uptodate_dir */ "Updating '%s' would lose untracked files in it",
@@ -34,7 +34,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' overlaps with '%s'. Cannot bind." ,/* sparse_not_uptodate_file */ -"Entry '%s' not uptodate. Cannot update sparse checkout." , +"Entry '%s' not up-to-date. Cannot update sparse checkout." ,/* would_lose_orphaned */ "Working tree file '%s' would be %s by sparse checkout update.",
Signature · 1 line
1.7.2.rc1.212.g4c287
Re: [PATCH v2] use "up-to-date" instead of "uptodate" for consistency
Quoted 10 lines · Nicolas Sebrecht
unpack-trees.c +1
-1@@ -22,7 +22,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' would be overwritten by merge. Cannot merge." ,/* not_uptodate_file */ -"Entry '%s' not uptodate. Cannot merge." , +"Entry '%s' not up-to-date. Cannot merge." ,/* not_uptodate_dir */ "Updating '%s' would lose untracked files in it",
unpack-trees.c static struct unpack_trees_error_msgs unpack_plumbing_errors = { · line 40
Quoted 8 lines
unpack-trees.c +1
-1@@ -34,7 +34,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {"Entry '%s' overlaps with '%s'. Cannot bind." ,/* sparse_not_uptodate_file */ -"Entry '%s' not uptodate. Cannot update sparse checkout." , +"Entry '%s' not up-to-date. Cannot update sparse checkout." ,/* would_lose_orphaned */ "Working tree file '%s' would be %s by sparse checkout update.",
What word do you see before "errors" on the hunk comment line?
[PATCH v2] Re: use "up-to-date" instead of "uptodate" for consistency
Quoted 5 lines · Junio C Hamano
Review static struct unpack_trees_error_msgs unpack_plumbing_errors = {
Quoted 1 line · Nicolas Sebrecht
(quoted diff) +0
-0@@ -22,7 +22,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {What word do you see before "errors" on the hunk comment line?
So you'd rather not touching it? Or wait until the next release cycle? Or a better commit message? Anything else?
Signature · 1 line
Nicolas Sebrecht
Re: [PATCH v2] Re: use "up-to-date" instead of "uptodate" for consistency
Quoted 8 lines · Nicolas Sebrecht
Quoted 5 lines · Junio C Hamano
Review static struct unpack_trees_error_msgs unpack_plumbing_errors = {
Quoted 1 line · Nicolas Sebrecht
(quoted diff) +0
-0@@ -22,7 +22,7 @@ static struct unpack_trees_error_msgs unpack_plumbing_errors = {What word do you see before "errors" on the hunk comment line?
So you'd rather not touching it?
Well, "'d rather" is actually too weak a statement.
The "struct unpack_trees_error_msgs" mechanism was introduced so that we can change the Porcelain level messages without breaking the plumbing API, which these messages are part of. Please see 8ccba00 (unpack-trees: allow Porcelain to give different error messages, 2008-05-17) and fadd069 (merge-recursive: give less scary messages when merge did not start, 2009-09-07) for backstory.
PATCH v3 1/1 merge-recursive: use "up-to-date" instead of "uptodate" in error message for consistency
---
Quoted 6 lines · Junio C Hamano
The "struct
unpack_trees_error_msgs" mechanism was introduced so that we can change the Porcelain level messages without breaking the plumbing API, which these messages are part of. Please see 8ccba00 (unpack-trees: allow Porcelain to give different error messages, 2008-05-17) and fadd069 (merge-recursive: give less scary messages when merge did not start, 2009-09-07) for backstory.
Oh, thanks. I was unaware of this mechanism and didn't care about what I was touching.
Documentation/git-checkout.txt | 2 +- merge-recursive.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-)
Documentation/git-checkout.txt +1 -1
@@ -263,7 +263,7 @@ the above checkout would fail like this: + ------------ $ git checkout mytopic -fatal: Entry 'frotz' not uptodate. Cannot merge. +fatal: Entry 'frotz' not up-to-date. Cannot merge. ------------ + You can give the `-m` flag to the command, which would try a
merge-recursive.c +1 -1
@@ -1214,7 +1214,7 @@ int merge_trees(struct merge_options *o, }if (sha_eq (common->object.sha1, merge->object.sha1)) { -output (o,0 ,"Already uptodate!" ); +output (o,0 ,"Already up-to-date!" ); *result = head;return 1 ; }
Signature · 1 line
1.7.2.rc1.214.g95a9e
Re: [PATCH v3] merge-recursive: use "up-to-date" instead of "uptodate" in error message for consistency
Quoted 12 lines · Nicolas Sebrecht
Signed-off-by Nicolas Sebrecht <[email protected]>---
Quoted 6 lines · Junio C Hamano
The "struct
unpack_trees_error_msgs" mechanism was introduced so that we can change the Porcelain level messages without breaking the plumbing API, which these messages are part of. Please see 8ccba00 (unpack-trees: allow Porcelain to give different error messages, 2008-05-17) and fadd069 (merge-recursive: give less scary messages when merge did not start, 2009-09-07) for backstory.
Quoted 2 lines
Oh, thanks. I was unaware of this mechanism and didn't care about what I was touching.
I suspect that the documentation patch talks about a nonexistent reality. See 8ccba00 again ;-)
PATCH 1/2 checkout: accord documentation to what git does
---
Quoted 13 lines · Junio C Hamano
Quoted 12 lines · Nicolas Sebrecht
Signed-off-by Nicolas Sebrecht <[email protected]>---
Quoted 6 lines · Junio C Hamano
The "struct
unpack_trees_error_msgs" mechanism was introduced so that we can change the Porcelain level messages without breaking the plumbing API, which these messages are part of. Please see 8ccba00 (unpack-trees: allow Porcelain to give different error messages, 2008-05-17) and fadd069 (merge-recursive: give less scary messages when merge did not start, 2009-09-07) for backstory.
Quoted 5 lines
Quoted 2 lines
Oh, thanks. I was unaware of this mechanism and didn't care about what I was touching.
I suspect that the documentation patch talks about a nonexistent reality. See 8ccba00 again ;-)
Doh! You said in this commit that
"If you do not see something wrong with this output, your brain has already been rotten with use of git for too long a time."
but I claim the right to be rotten in much more ways. :-)
Documentation/git-checkout.txt | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
Documentation/git-checkout.txt +1 -1
@@ -263,7 +263,7 @@ the above checkout would fail like this: + ------------ $ git checkout mytopic -fatal: Entry 'frotz' not uptodate. Cannot merge. +error: You have local changes to 'frotz'; not switching branches. ------------ + You can give the `-m` flag to the command, which would try a
Signature · 1 line
1.7.2.rc2.194.g494e9.dirty
Re: [PATCH 1/2] checkout: accord documentation to what git does
Quoted 6 lines · Nicolas Sebrecht
Doh! You said in this commit that
"If you do not see something wrong with this output, your brain has already been rotten with use of git for too long a time."
but I claim the right to be rotten in much more ways. :-)
Heh; thanks.
PATCH v4 2/2 merge-recursive: use "up-to-date" instead of "uptodate" in error message for consistency
In reply to Junio C Hamano--- merge-recursive.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
merge-recursive.c +1 -1
@@ -1214,7 +1214,7 @@ int merge_trees(struct merge_options *o, }if (sha_eq (common->object.sha1, merge->object.sha1)) { -output (o,0 ,"Already uptodate!" ); +output (o,0 ,"Already up-to-date!" ); *result = head;return 1 ; }
Signature · 1 line
1.7.2.rc2.194.g494e9.dirty
Re: [PATCH v4 2/2] merge-recursive: use "up-to-date" instead of "uptodate" in error message for consistency
Quoted 10 lines · Nicolas Sebrecht
Signed-off-by Nicolas Sebrecht <[email protected]>--- merge-recursive.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)merge-recursive.c +0
-0
merge-recursive.c int merge_trees(struct merge_options *o, · line 1220
Quoted 8 lines
merge-recursive.c +1
-1@@ -1214,7 +1214,7 @@ int merge_trees(struct merge_options *o, }if (sha_eq (common->object.sha1, merge->object.sha1)) { -output (o,0 ,"Already uptodate!" ); +output (o,0 ,"Already up-to-date!" ); *result = head;return 1 ; }
Sorry, but the pros-and-cons is not good enough for me to support this change. It will break users who have been parsing output with scripts.
We do strongly warn people against relying on Porcelain output, but that does not mean we are free to change them without having a good reason or two. The new spelling won't help the next reader who will find the message confusing the same way you did a few days ago.
IOW, if we are touching this line anyway, I'd like to make sure we made an effort to make it less confusing, not just spelled correctly, while we still have our memory of confusion fresh ;-)