HEAD's reflog entry for a renamed branch
Hello,
I noticed that, after renaming the current branch, the corresponding message in .git/logs/HEAD is empty.
For example, running
$ mkdir test-repo
$ cd test-repo
$ git init
$ echo abc >file.txt
$ git add file.txt
$ git commit -m"Add file.txt"
$ git branch -m master new-masterresulted in the following reflogs:
$ cat .git/logs/refs/heads/new-master 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 Branch: renamed refs/heads/master to refs/heads/new-master
$ cat .git/logs/HEAD 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 00000... Kyle Meyer <[email protected]> 1484607020 -0500
I expected the second line of .git/logs/HEAD to mirror the second line of .git/logs/refs/heads/new-master. Are the empty message and null sha1 in HEAD's entry intentional?
--
Kyle
Quoted 5 lines · Kyle Meyer
I noticed that, after renaming the current branch, the corresponding message in .git/logs/HEAD is empty.
For example, running
Quoted 7 lines
$ mkdir test-repo
$ cd test-repo
$ git init
$ echo abc >file.txt
$ git add file.txt
$ git commit -m"Add file.txt"
$ git branch -m master new-masterThanks for providing a simple reproduction recipe.
Quoted 6 lines
resulted in the following reflogs:
$ cat .git/logs/refs/heads/new-master 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 Branch: renamed refs/heads/master to refs/heads/new-master
Quoted 7 lines
$ cat .git/logs/HEAD 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 00000... Kyle Meyer <[email protected]> 1484607020 -0500I expected the second line of .git/logs/HEAD to mirror the second line of .git/logs/refs/heads/new-master. Are the empty message and null sha1 in HEAD's entry intentional?
The null sha1 is correct, I think. The branch we were on went away, and we use the null sha1 to indicate "no value" in both the creation and deletion cases.
It's unfortunate that there's no message. This is because the rename calls delete_ref() under the hood, but that function doesn't take a reflog message argument at all. It usually doesn't matter because deleting the ref will also delete the reflog.
But as your example shows, deletions _can_ get logged in the HEAD reflog; the low-level ref code detects when HEAD points to the updated ref and logs an entry there, too.
You can see the same thing with a simpler example:
$ git init $ git commit -m foo --allow-empty $ git update-ref -d refs/heads/master $ cat .git/logs/HEAD 00000... 8ffd1... Jeff King <[email protected]> 1485464779 -0500 commit (initial): foo 8ffd1... 00000... Jeff King <[email protected]> 1485464787 -0500
This doesn't come up much because most porcelain commands will refuse to delete the current branch (notice I had to use "update-ref" and not "branch -d").
I'd say there are two potential improvements:
- delete_ref() should take a reflog message argument, in case it updates the HEAD reflog (as a bonus, this will future-proof us for a day when we might keep reflogs for deleted refs).
- "git branch -m" does seem to realize when we are renaming HEAD, because it updates HEAD to point to the new branch name. But it should probably insert another reflog entry mentioning the rename (we do for "git checkout foo", even when "foo" has the same sha1 as the current HEAD).
-Peff
Quoted 11 lines · Jeff King
It's unfortunate that there's no message. This is because the rename calls
delete_ref()under the hood, but that function doesn't take a reflog message argument at all. It usually doesn't matter because deleting the ref will also delete the reflog.But as your example shows, deletions _can_ get logged in the HEAD reflog; the low-level ref code detects when HEAD points to the updated ref and logs an entry there, too.
...
I'd say there are two potential improvements:
Quoted 3 lines
-
delete_ref()should take a reflog message argument, in case it updates the HEAD reflog (as a bonus, this will future-proof us for a day when we might keep reflogs for deleted refs).
This sounds sensible.
Quoted 5 lines
- "git branch -m" does seem to realize when we are renaming HEAD, because it updates HEAD to point to the new branch name. But it should probably insert another reflog entry mentioning the rename (we do for "git checkout foo", even when "foo" has the same sha1 as the current HEAD).
This one I care less (not in the sense that I prefer it not done, but in the sense that I do not mind it is left unfixed than the other one you pointed out).
Quoted 9 lines · Junio C Hamano
Quoted 5 lines
- "git branch -m" does seem to realize when we are renaming HEAD, because it updates HEAD to point to the new branch name. But it should probably insert another reflog entry mentioning the rename (we do for "git checkout foo", even when "foo" has the same sha1 as the current HEAD).
This one I care less (not in the sense that I prefer it not done, but in the sense that I do not mind it is left unfixed than the other one you pointed out).
I wondered if it might affect how "git checkout -" works. But that feature looks for reflogs like "checkout: moving from X to Y" to know to move back to X. So we are fine here. Even though the HEAD reflog does not show us going _to_ new-master, we would see it in a later entry as "from new-master to Y". What we are missing is "rename from master to new-master", but that entry does not matter. There is no "master" to go back to anymore. :)
-Peff
cover delete_ref(): support reflog messages
In reply to Jeff King[Sorry for the slow response.]
Quoted 1 line · Jeff King
On Mon, Jan 16, 2017 at 06:17:29PM -0500, Kyle Meyer wrote:
[...]
Quoted 8 lines
Quoted 8 lines
$ cat .git/logs/refs/heads/new-master 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 Branch: renamed refs/heads/master to refs/heads/new-master$ cat .git/logs/HEAD 00000... 68730... Kyle Meyer <[email protected]> 1484607020 -0500 commit (initial): Add file.txt 68730... 00000... Kyle Meyer <[email protected]> 1484607020 -0500
Quoted 7 lines
Quoted 3 lines
I expected the second line of .git/logs/HEAD to mirror the second line of .git/logs/refs/heads/new-master. Are the empty message and null sha1 in HEAD's entry intentional?
The null sha1 is correct, I think. The branch we were on went away, and we use the null sha1 to indicate "no value" in both the creation and deletion cases.
I see, thanks.
Quoted 5 lines
I'd say there are two potential improvements:
-
delete_ref()should take a reflog message argument, in case it updates the HEAD reflog (as a bonus, this will future-proof us for a day when we might keep reflogs for deleted refs).
I've tried to do this in the following patches.
Quoted 5 lines
- "git branch -m" does seem to realize when we are renaming HEAD, because it updates HEAD to point to the new branch name. But it should probably insert another reflog entry mentioning the rename (we do for "git checkout foo", even when "foo" has the same sha1 as the current HEAD).
I haven't worked out how to do this part yet. I'm guessing the change will involve modifying split_head_update().
If this is added, should it be instead of, rather than in addition to, the deletion entry? If a "Branch: renamed ..." entry is present, it doesn't seem like the deletion entry is providing any additional information.
delete_refs(): accept a reflog message argument
update-ref: pass reflog message argument to delete_refs
rename_ref: replace empty deletion message in HEAD's log
builtin/am.c | 4 ++-- builtin/branch.c | 2 +- builtin/notes.c | 4 ++-- builtin/remote.c | 4 ++-- builtin/replace.c | 2 +- builtin/reset.c | 2 +- builtin/symbolic-ref.c | 2 +- builtin/tag.c | 2 +- builtin/update-ref.c | 2 +- fast-import.c | 2 +- refs.c | 4 ++-- refs.h | 2 +- refs/files-backend.c | 12 +++++++++--- t/t1400-update-ref.sh | 9 +++++++++ t/t3200-branch.sh | 4 ++++ transport.c | 2 +- 16 files changed, 39 insertions(+), 20 deletions(-)
Signature · 1 line
Kyle
PATCH 1/3 delete_refs(): accept a reflog message argument
When the current branch is renamed with 'git branch -m/-M' or deleted with 'git update-ref -m<msg> -d', the event is recorded in HEAD's log with an empty message.
In preparation for adding a more meaningful message to HEAD's log in these cases, update delete_ref() to take a message argument and pass it along to ref_transaction_delete(). Modify all callers to pass NULL for the new message argument; no change in behavior is intended.
--- builtin/am.c | 4 ++-- builtin/branch.c | 2 +- builtin/notes.c | 4 ++-- builtin/remote.c | 4 ++-- builtin/replace.c | 2 +- builtin/reset.c | 2 +- builtin/symbolic-ref.c | 2 +- builtin/tag.c | 2 +- builtin/update-ref.c | 2 +- fast-import.c | 2 +- refs.c | 4 ++-- refs.h | 2 +- refs/files-backend.c | 6 +++--- transport.c | 2 +- 14 files changed, 20 insertions(+), 20 deletions(-)
builtin/am.c +2 -2
@@ -1049,7 +1049,7 @@ static void am_setup(struct am_state *state, enum patch_format patch_format, }else {write_state_text (state,"abort-safety" ,"" );if (!state->rebasing) -delete_ref ("ORIG_HEAD" ,NULL ,0 ); +delete_ref ("ORIG_HEAD" ,NULL ,0 ,NULL ); }/*
@@ -2172,7 +2172,7 @@ static void am_abort(struct am_state *state) has_curr_head ? &curr_head :NULL ,0 , UPDATE_REFS_DIE_ON_ERR); else if (curr_branch) - delete_ref(curr_branch, NULL, REF_NODEREF); + delete_ref(curr_branch, NULL, REF_NODEREF, NULL);free (curr_branch );am_destroy (state);
builtin/branch.c +1 -1
@@ -252,7 +252,7 @@ static int delete_branches(int argc, const char **argv, int force, int kinds, }if (delete_ref (name,is_null_sha1 (sha1) ?NULL : sha1, -REF_NODEREF )) { +REF_NODEREF ,NULL )) { error(remote_branch ? _("Error deleting remote-tracking branch '%s'") : _("Error deleting branch '%s'"),
builtin/notes.c +2 -2
@@ -681,9 +681,9 @@ static int merge_abort(struct notes_merge_options *o)* notes_merge_abort() to remove .git/NOTES_MERGE_WORKTREE. */ -if (delete_ref ("NOTES_MERGE_PARTIAL" ,NULL ,0 )) +if (delete_ref ("NOTES_MERGE_PARTIAL" ,NULL ,0 ,NULL )) ret +=error (_ ("failed to delete ref NOTES_MERGE_PARTIAL" )); -if (delete_ref ("NOTES_MERGE_REF" ,NULL ,REF_NODEREF )) +if (delete_ref ("NOTES_MERGE_REF" ,NULL ,REF_NODEREF ,NULL )) ret +=error (_ ("failed to delete ref NOTES_MERGE_REF" ));if (notes_merge_abort (o)) ret +=error (_ ("failed to remove 'git notes merge' worktree" ));
builtin/remote.c +2 -2
@@ -691,7 +691,7 @@ static int mv(int argc, const char **argv)read_ref_full (item->string,RESOLVE_REF_READING , oid.hash, &flag);if (!(flag &REF_ISSYMREF ))continue ; -if (delete_ref (item->string,NULL ,REF_NODEREF )) +if (delete_ref (item->string,NULL ,REF_NODEREF ,NULL ))die (_ ("deleting '%s' failed" ), item->string); }for (i =0 ; i < remote_branches.nr; i++) {
@@ -1248,7 +1248,7 @@ static int set_head(int argc, const char **argv) head_name =xstrdup (states.heads.items[0 ].string);free_remote_ref_states (&states); }else if (opt_d && !opt_a && argc ==1 ) { -if (delete_ref (buf.buf,NULL ,REF_NODEREF )) +if (delete_ref (buf.buf,NULL ,REF_NODEREF ,NULL )) result |=error (_ ("Could not delete %s" ), buf.buf); }else usage_with_options (builtin_remote_sethead_usage, options);
builtin/replace.c +1 -1
@@ -121,7 +121,7 @@ static int for_each_replace_name(const char **argv, each_replace_name_fn fn)static int delete_replace_ref (const char *name,const char *ref,const unsigned char *sha1) { -if (delete_ref (ref, sha1,0 )) +if (delete_ref (ref, sha1,0 ,NULL ))return 1 ;printf ("Deleted replace ref '%s'\n" , name);return 0 ;
builtin/reset.c +1 -1
@@ -256,7 +256,7 @@ static int reset_refs(const char *rev, const struct object_id *oid)update_ref_oid (msg.buf,"ORIG_HEAD" , orig, old_orig,0 ,UPDATE_REFS_MSG_ON_ERR ); }else if (old_orig) -delete_ref ("ORIG_HEAD" , old_orig->hash,0 ); +delete_ref ("ORIG_HEAD" , old_orig->hash,0 ,NULL );set_reflog_message (&msg,"updating HEAD" , rev); update_ref_status =update_ref_oid (msg.buf,"HEAD" , oid, orig,0 ,UPDATE_REFS_MSG_ON_ERR );
builtin/symbolic-ref.c +1 -1
@@ -58,7 +58,7 @@ int cmd_symbolic_ref(int argc, const char **argv, const char *prefix)die ("Cannot delete %s, not a symbolic ref" , argv[0 ]);if (!strcmp (argv[0 ],"HEAD" ))die ("deleting '%s' is not allowed" , argv[0 ]); -return delete_ref (argv[0 ],NULL ,REF_NODEREF ); +return delete_ref (argv[0 ],NULL ,REF_NODEREF ,NULL ); }switch (argc) {
builtin/tag.c +1 -1
@@ -97,7 +97,7 @@ static int for_each_tag_name(const char **argv, each_tag_name_fn fn,static int delete_tag (const char *name,const char *ref,const unsigned char *sha1,const void *cb_data) { -if (delete_ref (ref, sha1,0 )) +if (delete_ref (ref, sha1,0 ,NULL ))return 1 ;printf (_ ("Deleted tag '%s' (was %s)\n" ), name,find_unique_abbrev (sha1,DEFAULT_ABBREV ));return 0 ;
builtin/update-ref.c +1 -1
@@ -435,7 +435,7 @@ int cmd_update_ref(int argc, const char **argv, const char *prefix)*/ return delete_ref (refname, (oldval && !is_null_sha1 (oldsha1)) ? oldsha1 :NULL , - flags); + flags,NULL ); else return update_ref(msg, refname, sha1, oldval ? oldsha1 : NULL, flags | create_reflog_flag,
fast-import.c +1 -1
@@ -1752,7 +1752,7 @@ static int update_branch(struct branch *b)if (is_null_sha1 (b->sha1)) {if (b->delete) -delete_ref (b->name,NULL ,0 ); +delete_ref (b->name,NULL ,0 ,NULL );return 0 ; }if (read_ref (b->name, old_sha1))
refs.c +2 -2
@@ -592,7 +592,7 @@ static int delete_pseudoref(const char *pseudoref, const unsigned char *old_sha1 }int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags) +unsigned int flags,const char *msg) {struct ref_transaction *transaction;struct strbuf err =STRBUF_INIT ;
@@ -603,7 +603,7 @@ int delete_ref(const char *refname, const unsigned char *old_sha1, transaction =ref_transaction_begin (&err);if (!transaction ||ref_transaction_delete (transaction, refname, old_sha1, - flags,NULL , &err) || + flags, msg, &err) ||ref_transaction_commit (transaction, &err)) {error ("%s" , err.buf);ref_transaction_free (transaction);
refs.h +1 -1
@@ -277,7 +277,7 @@ int reflog_exists(const char *refname);* be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +unsigned int flags,const char *msg);/* * Delete the specified references. If there are any problems, emit
refs/files-backend.c +3 -3
@@ -2489,7 +2489,7 @@ static int files_delete_refs(struct ref_store *ref_store,for (i =0 ; i < refnames->nr; i++) {const char *refname = refnames->items[i].string; -if (delete_ref (refname,NULL , flags)) +if (delete_ref (refname,NULL , flags,NULL )) result |=error (_ ("could not remove reference %s" ), refname); }
@@ -2616,7 +2616,7 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (oldrefname, orig_sha1,REF_NODEREF )) { +if (delete_ref (oldrefname, orig_sha1,REF_NODEREF ,NULL )) {error ("unable to delete old %s" , oldrefname); goto rollback; }
@@ -2630,7 +2630,7 @@ static int files_rename_ref(struct ref_store *ref_store,*/ if (!read_ref_full (newrefname,RESOLVE_REF_READING |RESOLVE_REF_NO_RECURSE , sha1,NULL ) && -delete_ref (newrefname,NULL ,REF_NODEREF )) { +delete_ref (newrefname,NULL ,REF_NODEREF ,NULL )) {if (errno==EISDIR ) {struct strbuf path =STRBUF_INIT ;int result;
transport.c +1 -1
@@ -299,7 +299,7 @@ void transport_update_tracking_ref(struct remote *remote, struct ref *ref, int vif (verbose)fprintf (stderr,"updating local tracking ref '%s'\n" , rs.dst);if (ref->deletion) { -delete_ref (rs.dst,NULL ,0 ); +delete_ref (rs.dst,NULL ,0 ,NULL ); }else update_ref ("update by push" , rs.dst, ref->new_oid.hash,NULL ,0 ,0 );
Signature · 1 line
2.11.1
PATCH 2/3 update-ref: pass reflog message argument to delete_refs
In reply to Kyle MeyerNow that delete_refs() accepts a reflog message, pass the user-provided message to delete_refs() rather than silently dropping it. The doesn't matter for the deleted ref's log because the log is deleted along with the ref, but this entry will show up in HEAD's reflog when deleting a checked out branch.
---
I'm not sure if the test here (or in the next patch) is worth including.
builtin/update-ref.c | 2 +- t/t1400-update-ref.sh | 9 +++++++++ 2 files changed, 10 insertions(+), 1 deletion(-)
builtin/update-ref.c +1 -1
@@ -435,7 +435,7 @@ int cmd_update_ref(int argc, const char **argv, const char *prefix)*/ return delete_ref (refname, (oldval && !is_null_sha1 (oldsha1)) ? oldsha1 :NULL , - flags,NULL ); + flags, msg); else return update_ref(msg, refname, sha1, oldval ? oldsha1 : NULL, flags | create_reflog_flag,
t/t1400-update-ref.sh +9 -0
@@ -85,6 +85,15 @@ test_expect_success "delete $m (by HEAD)" ' ' rm -f .git/$m +test_expect_success "deleting current branch adds message to HEAD's log" ' + git update-ref $m $A && + git symbolic-ref HEAD $m && + git update-ref -mdelmsg -d $m && + ! test -f .git/$m && + grep "delmsg$" .git/logs/HEAD >/dev/null +' +rm -f .git/$m + test_expect_success 'update-ref does not create reflogs by default' ' test_when_finished "git update-ref -d $outside" && git update-ref $outside $A &&
Signature · 1 line
2.11.1
PATCH 3/3 rename_ref: replace empty deletion message in HEAD's log
In reply to Kyle MeyerWhen the current branch is renamed, the deletion of the old ref is recorded in HEAD's log with an empty message. Now that delete_refs() accepts a reflog message, provide a more descriptive message. This message will not be included in the reflog of the renamed branch, but its log already covers the renaming event with a message of "Branch:
renamed ...".
--- refs/files-backend.c | 8 +++++++- t/t3200-branch.sh | 4 ++++ 2 files changed, 11 insertions(+), 1 deletion(-)
refs/files-backend.c +7 -1
@@ -2598,6 +2598,7 @@ static int files_rename_ref(struct ref_store *ref_store,struct stat loginfo;int log = !lstat (git_path ("logs/%s" , oldrefname), &loginfo);struct strbuf err =STRBUF_INIT ; +struct strbuf logmsg_del =STRBUF_INIT ;if (log &&S_ISLNK (loginfo.st_mode))return error ("reflog for %s is a symlink" , oldrefname);
@@ -2616,10 +2617,15 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (oldrefname, orig_sha1,REF_NODEREF ,NULL )) { +strbuf_addf (&logmsg_del,"Deleted %s" , oldrefname); + +if (delete_ref (oldrefname, orig_sha1,REF_NODEREF , logmsg_del.buf)) {error ("unable to delete old %s" , oldrefname); +strbuf_release (&logmsg_del); goto rollback; } +strbuf_release (&logmsg_del); +/* * Since we are doing a shallow lookup, sha1 is not the
t/t3200-branch.sh +4 -0
@@ -139,6 +139,10 @@ test_expect_success 'git branch -M baz bam should succeed when baz is checked ou test $(git rev-parse --abbrev-ref HEAD) = bam ' +test_expect_success 'git branch -M baz bam should add entry to .git/logs/HEAD' ' + grep "Deleted refs/heads/baz$" .git/logs/HEAD >/dev/null +' + test_expect_success 'git branch -M baz bam should succeed when baz is checked out as linked working tree' ' git checkout master && git worktree add -b baz bazdir &&
Signature · 1 line
2.11.1
Re: [PATCH 1/3] delete_refs(): accept a reflog message argument
In reply to Kyle MeyerQuoted 8 lines · Kyle Meyer
When the current branch is renamed with 'git branch -m/-M' or deleted with 'git update-ref -m<msg> -d', the event is recorded in HEAD's log with an empty message.
In preparation for adding a more meaningful message to HEAD's log in these cases, update
delete_ref()to take a message argument and pass it along toref_transaction_delete(). Modify all callers to pass NULL for the new message argument; no change in behavior is intended.
Seems like a good first step.
refs.h int reflog_exists(const char *refname); · line 280
Quoted 10 lines
refs.h +1
-1@@ -277,7 +277,7 @@ int reflog_exists(const char *refname);* be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +unsigned int flags,const char *msg);
Should the "msg" argument go at the beginning, to match update_ref()?
-Peff
Re: [PATCH 0/3] delete_ref(): support reflog messages
In reply to Kyle MeyerQuoted 1 line · Kyle Meyer
[Sorry for the slow response.]
No problem. The pace of open source varies wildly. :)
Quoted 6 lines
Quoted 5 lines
- "git branch -m" does seem to realize when we are renaming HEAD, because it updates HEAD to point to the new branch name. But it should probably insert another reflog entry mentioning the rename (we do for "git checkout foo", even when "foo" has the same sha1 as the current HEAD).
Quoted 7 lines
I haven't worked out how to do this part yet. I'm guessing the change will involve modifying
split_head_update().If this is added, should it be instead of, rather than in addition to, the deletion entry? If a "Branch: renamed ..." entry is present, it doesn't seem like the deletion entry is providing any additional information.
I think you could do an "instead of" that goes from sha1 X to X, and just mentions the rename. Or you could add a second entry after the delete that takes it from 0{40} back to X.
I suspect the latter is easier to do, and I doubt anybody would care that much of the exact form. These entries aren't really doing anything for reachability. They're just giving an audit log of what happened. So I don't think anybody would really care unless they were debugging a confusing situation by hand. And as long as there's enough information to figure out what happened, they'll be happy.
-Peff
Re: [PATCH 2/3] update-ref: pass reflog message argument to delete_refs
In reply to Kyle MeyerQuoted 5 lines · Kyle Meyer
Now that
delete_refs()accepts a reflog message, pass the user-provided message todelete_refs()rather than silently dropping it. The doesn't matter for the deleted ref's log because the log is deleted along with the ref, but this entry will show up in HEAD's reflog when deleting a checked out branch.
Sounds good.
builtin/update-ref.c int cmd_update_ref(int argc, const char **argv, const char *prefix) · line 438
Quoted 10 lines
builtin/update-ref.c +1
-1@@ -435,7 +435,7 @@ int cmd_update_ref(int argc, const char **argv, const char *prefix)*/ return delete_ref (refname, (oldval && !is_null_sha1 (oldsha1)) ? oldsha1 :NULL , - flags,NULL ); + flags, msg);
This looks obviously correct.
Quoted 6 lines
t/t1400-update-ref.sh +0
-0@@ -85,6 +85,15 @@ test_expect_success "delete $m (by HEAD)" ' '
t/t1400-update-ref.sh test_expect_success "delete $m (by HEAD)" ' · line 95
Quoted 10 lines
t/t1400-update-ref.sh +8
-0rm -f .git/$m +test_expect_success "deleting current branch adds message to HEAD's log" ' + git update-ref $m $A && + git symbolic-ref HEAD $m && + git update-ref -mdelmsg -d $m && + ! test -f .git/$m && + grep "delmsg$" .git/logs/HEAD >/dev/null +' +rm -f .git/$m
I think covering this with a test is good.
I don't know if it's also worth testing that deleting via HEAD also writes the reflog. I.e.,:
git update-ref -m delete-by-head -d HEAD
Some of the style here is a bit out-dated, but I think you are just matching the surrounding tests. So that's OK by me (though a patch to modernize the whole thing would be welcome, too).
For reference, the two things I notice are:
- we prefer test_path_is_missing to "! test -f" these days.
- we don't redirect the output of grep (it's handled already in non-verbose mode, and in verbose mode we try to be...verbose).
-Peff
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Kyle MeyerQuoted 6 lines · Kyle Meyer
When the current branch is renamed, the deletion of the old ref is recorded in HEAD's log with an empty message. Now that
delete_refs()accepts a reflog message, provide a more descriptive message. This message will not be included in the reflog of the renamed branch, but its log already covers the renaming event with a message of "Branch:
renamed ...".
Right, makes sense. The code overall looks fine, though I have one nit:
refs/files-backend.c static int files_rename_ref(struct ref_store *ref_store, · line 2620
Quoted 6 lines
refs/files-backend.c +1
-1@@ -2616,10 +2617,15 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (oldrefname, orig_sha1,REF_NODEREF ,NULL )) { +strbuf_addf (&logmsg_del,"Deleted %s" , oldrefname);
We've been anything but consistent with our reflog messages in the past, but I think the general guideline for new ones is to use "command:
action". Of course we don't _know_ the action here, because we're deep in rename_ref().
Should we have the caller pass it in for us, as we do with delete_ref() and update_ref()?
I see we actually already have a "logmsg" parameter. It already says "Branch: renamed %s to %s". Could we just reuse that? I know that this step is technically just the deletion, but I think it more accurately describes the whole operation that the deletion is part of.
-Peff
Re: [PATCH 1/3] delete_refs(): accept a reflog message argument
In reply to Jeff KingQuoted 4 lines · Jeff King
Quoted 4 lines
refs.h +0
-0
Quoted 8 lines
refs.h int reflog_exists(const char *refname); · line 280
Quoted 6 lines
refs.h +1
-1@@ -277,7 +277,7 @@ int reflog_exists(const char *refname);* be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +unsigned int flags,const char *msg);Should the "msg" argument go at the beginning, to match
update_ref()?
Probably. rename/create have the message at the end but their parameters are very different from update/delete. The parameters update and delete take are not identical, but we can view them as a lot more similar than the other two. So I think it makes sense for delete to try matching update, even though trying to make all four the same may proabably be pointless.
Re: [PATCH 2/3] update-ref: pass reflog message argument to delete_refs
In reply to Jeff KingQuoted 1 line · Jeff King
...
All good review comments.
I briefly wondered if recording the deletion of the current branch in HEAD's reflog has much practical values (Porcelain "git branch" would not even allow deletion in the first place), but because it is a rare and unusual event, it probably makes more sense to have a record of it.
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Jeff KingQuoted 24 lines · Jeff King
Quoted 6 lines · Kyle Meyer
When the current branch is renamed, the deletion of the old ref is recorded in HEAD's log with an empty message. Now that
delete_refs()accepts a reflog message, provide a more descriptive message. This message will not be included in the reflog of the renamed branch, but its log already covers the renaming event with a message of "Branch:
renamed ...".Right, makes sense. The code overall looks fine, though I have one nit:
refs/files-backend.c static int files_rename_ref(struct ref_store *ref_store, · line 2620
Quoted 6 lines
refs/files-backend.c +1
-1@@ -2616,10 +2617,15 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (oldrefname, orig_sha1,REF_NODEREF ,NULL )) { +strbuf_addf (&logmsg_del,"Deleted %s" , oldrefname);We've been anything but consistent with our reflog messages in the past, but I think the general guideline for new ones is to use "command:
action". Of course we don't _know_ the action here, because we're deep inrename_ref().
Quoted 7 lines
Should we have the caller pass it in for us, as we do with
delete_ref()andupdate_ref()?I see we actually already have a "logmsg" parameter. It already says "Branch: renamed %s to %s". Could we just reuse that? I know that this step is technically just the deletion, but I think it more accurately describes the whole operation that the deletion is part of.
True, but stepping back a bit,...
Do we even want these "internal" delete_ref() invocations to be logged in HEAD's reflog? I understand that this is inside the implementation of renaming an old ref to a new ref, and reflog message given to delete_ref() would matter only if the HEAD happens to be pointing at old ref---but then HEAD will be repointed to the new ref by somebody else [*1*] that called this function to rename old to new and it _will_ log it. So I am not sure if it is a good thing to describe the deletion more readably with a message (which is what this patch does) in the first place. If we can just say "don't log this deletion event in HEAD's reflog", wouldn't that be more desirable?
[Footnote]
*1* Is the reason why the code in files_rename_ref() we are looking
at does not adjust HEAD to point at the new ref is because it is just handing one ref-store and obviouvious to symrefs in other backends?
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
Quoted 7 lines · Junio C Hamano
Quoted 4 lines
I see we actually already have a "logmsg" parameter. It already says "Branch: renamed %s to %s". Could we just reuse that? I know that this step is technically just the deletion, but I think it more accurately describes the whole operation that the deletion is part of.
True, but stepping back a bit,...
Quoted 11 lines
Do we even want these "internal"
delete_ref()invocations to be logged in HEAD's reflog? I understand that this is inside the implementation of renaming an old ref to a new ref, and reflog message given todelete_ref()would matter only if the HEAD happens to be pointing at old ref---but then HEAD will be repointed to the new ref by somebody else [*1*] that called this function to rename old to new and it _will_ log it. So I am not sure if it is a good thing to describe the deletion more readably with a message (which is what this patch does) in the first place. If we can just say "don't log this deletion event in HEAD's reflog", wouldn't that be more desirable?
Yes. I think the options are basically (in order of decreasing preference in my opinion):
1. Log a rename entry (same sha1, but note the rename in the free-form
text).
2. Log a delete (sha1 goes to null) followed by a creation (from null back to the original sha1).
3. Log nothing at all for HEAD.
This does half of (2). If we do the second half, then I'd prefer it to (3). But if we can do (1), that is better still (IMHO).
Quoted 4 lines
*1* Is the reason why the code in
files_rename_ref()we are looking
at does not adjust HEAD to point at the new ref is because it is just handing one ref-store and obviouvious to symrefs in other backends?
I'm actually confused about which bit of code is updating HEAD. I do not see it either in files_rename_ref() or in the caller. Yet it clearly happens. But that is the code that would know enough to do (1) or the second half of (2) above.
-Peff
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
Quoted 19 lines · Jeff King
Yes. I think the options are basically (in order of decreasing preference in my opinion):
1. Log a rename entry (same sha1, but note the rename in the free-form
text).2. Log a delete (sha1 goes to null) followed by a creation (from null back to the original sha1).
3. Log nothing at all for HEAD.
This does half of (2). If we do the second half, then I'd prefer it to (3). But if we can do (1), that is better still (IMHO).
Quoted 4 lines
*1* Is the reason why the code in
files_rename_ref()we are looking
at does not adjust HEAD to point at the new ref is because it is just handing one ref-store and obviouvious to symrefs in other backends?
Quoted 4 lines
I'm actually confused about which bit of code is updating HEAD. I do not see it either in
files_rename_ref()or in the caller. Yet it clearly happens. But that is the code that would know enough to do (1) or the second half of (2) above.
Ah, I found it. It's in replace_each_worktree_head_symref() these days, which does not bother to pass a log message.
So I think the second half of (2) is probably something like the patch below.
Thinking on it more, we probably _do_ want two entries. Because the operations are not atomic, it's possible that we may end up in a half-way state after the first entry is written. And when debugging such a case, I'd much rather see the first half of the operation logged than nothing at all.
-Peff
---
branch.c +3 -2
@@ -345,7 +345,8 @@ void die_if_checked_out(const char *branch, int ignore_current_worktree) branch, wt->path); } -int replace_each_worktree_head_symref (const char *oldref,const char *newref) +int replace_each_worktree_head_symref (const char *oldref,const char *newref, +const char *logmsg) {int ret =0 ;struct worktree **worktrees =get_worktrees (0 );
@@ -358,7 +359,7 @@ int replace_each_worktree_head_symref(const char *oldref, const char *newref)continue ;if (set_worktree_head_symref (get_worktree_git_dir (worktrees[i]), - newref)) { + newref, logmsg)) { ret =-1 ;error (_ ("HEAD of working tree %s is not updated" ), worktrees[i]->path);
branch.h +2 -1
@@ -71,6 +71,7 @@ extern void die_if_checked_out(const char *branch, int ignore_current_worktree);* This will be used when renaming a branch. Returns 0 if successful, non-zero * otherwise. */ -extern int replace_each_worktree_head_symref (const char *oldref,const char *newref); +extern int replace_each_worktree_head_symref (const char *oldref,const char *newref, +const char *logmsg);#endif
builtin/branch.c +3 -2
@@ -471,14 +471,15 @@ static void rename_branch(const char *oldname, const char *newname, int force)if (rename_ref (oldref.buf, newref.buf, logmsg.buf))die (_ ("Branch rename failed" )); -strbuf_release (&logmsg);if (recovery)warning (_ ("Renamed a misnamed branch '%s' away" ), oldref.buf +11 ); -if (replace_each_worktree_head_symref (oldref.buf, newref.buf)) +if (replace_each_worktree_head_symref (oldref.buf, newref.buf, logmsg.buf))die (_ ("Branch renamed to %s, but HEAD is not updated!" ), newname); +strbuf_release (&logmsg); +strbuf_addf (&oldsection,"branch.%s" , oldref.buf +11 );strbuf_release (&oldref);strbuf_addf (&newsection,"branch.%s" , newref.buf +11 );
refs.h +2 -1
@@ -334,7 +334,8 @@ int create_symref(const char *refname, const char *target, const char *logmsg);* $GIT_DIR points to. * Return 0 if successful, non-zero otherwise. * */ -int set_worktree_head_symref (const char *gitdir,const char *target); +int set_worktree_head_symref (const char *gitdir,const char *target, +const char *logmsg);enum action_on_err {UPDATE_REFS_MSG_ON_ERR ,
refs/files-backend.c +2 -2
@@ -3024,7 +3024,7 @@ static int files_create_symref(struct ref_store *ref_store,return ret; } -int set_worktree_head_symref (const char *gitdir,const char *target) +int set_worktree_head_symref (const char *gitdir,const char *target,const char *logmsg) {static struct lock_file head_lock;struct ref_lock *lock;
@@ -3052,7 +3052,7 @@ int set_worktree_head_symref(const char *gitdir, const char *target) lock->lk = &head_lock; lock->ref_name =xstrdup (head_rel); - ret =create_symref_locked (lock, head_rel, target,NULL ); + ret =create_symref_locked (lock, head_rel, target, logmsg);unlock_ref (lock);/* will free lock */ strbuf_release (&head_path);
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
Quoted 5 lines · Jeff King
Thinking on it more, we probably _do_ want two entries. Because the operations are not atomic, it's possible that we may end up in a half-way state after the first entry is written. And when debugging such a case, I'd much rather see the first half of the operation logged than nothing at all.
Yes, that sounds right.
Re: [PATCH 1/3] delete_refs(): accept a reflog message argument
In reply to Junio C HamanoQuoted 15 lines · Junio C Hamano
Quoted 4 lines · Jeff King
Quoted 4 lines
refs.h +0
-0Quoted 8 lines
refs.h int reflog_exists(const char *refname); · line 280
Quoted 6 lines
refs.h +1
-1@@ -277,7 +277,7 @@ int reflog_exists(const char *refname);* be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +unsigned int flags,const char *msg);Should the "msg" argument go at the beginning, to match
update_ref()?
Quoted 6 lines
Probably. rename/create have the message at the end but their parameters are very different from update/delete. The parameters update and delete take are not identical, but we can view them as a lot more similar than the other two. So I think it makes sense for delete to try matching update, even though trying to make all four the same may proabably be pointless.
I put "msg" after "flags" because that's where it occurs in ref_transaction_delete(), but matching update_ref() makes sense.
Signature · 1 line
Kyle
Re: [PATCH 2/3] update-ref: pass reflog message argument to delete_refs
In reply to Jeff KingJeff King <[email protected]> writes:
[...]
Quoted 19 lines
Quoted 6 lines
t/t1400-update-ref.sh +0
-0@@ -85,6 +85,15 @@ test_expect_success "delete $m (by HEAD)" ' 't/t1400-update-ref.sh test_expect_success "delete $m (by HEAD)" ' · line 95
Quoted 10 lines
t/t1400-update-ref.sh +8
-0rm -f .git/$m +test_expect_success "deleting current branch adds message to HEAD's log" ' + git update-ref $m $A && + git symbolic-ref HEAD $m && + git update-ref -mdelmsg -d $m && + ! test -f .git/$m && + grep "delmsg$" .git/logs/HEAD >/dev/null +' +rm -f .git/$mI think covering this with a test is good.
Quoted 4 lines
I don't know if it's also worth testing that deleting via HEAD also writes the reflog. I.e.,:
git update-ref -m delete-by-head -d HEAD
Seems reasonable to cover this case as well.
Quoted 3 lines
Some of the style here is a bit out-dated, but I think you are just matching the surrounding tests. So that's OK by me (though a patch to modernize the whole thing would be welcome, too).
Right. I'd be happy to follow up with a patch updating the style in t1400-update-ref.sh.
Quoted 6 lines
For reference, the two things I notice are:
- we prefer
test_path_is_missingto "! test -f" these days.- we don't redirect the output of grep (it's handled already in non-verbose mode, and in verbose mode we try to be...verbose).
Would moving cleanup like "rm -f .git/$m" within the test's body using test_when_finished also be preferred?
Signature · 1 line
Kyle
Re: [PATCH 2/3] update-ref: pass reflog message argument to delete_refs
Quoted 9 lines · Kyle Meyer
Quoted 6 lines
For reference, the two things I notice are:
- we prefer
test_path_is_missingto "! test -f" these days.- we don't redirect the output of grep (it's handled already in non-verbose mode, and in verbose mode we try to be...verbose).
Would moving cleanup like "rm -f .git/$m" within the test's body using
test_when_finishedalso be preferred?
Yeah, that too. I forgot to mention it.
-Peff
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Junio C HamanoJunio C Hamano <[email protected]> writes:
[...]
Quoted 7 lines
Do we even want these "internal"
delete_ref()invocations to be logged in HEAD's reflog? I understand that this is inside the implementation of renaming an old ref to a new ref, and reflog message given todelete_ref()would matter only if the HEAD happens to be pointing at old ref---but then HEAD will be repointed to the new ref by somebody else [*1*] that called this function to rename old to new and it _will_ log it.
I know the discussion has developed further, but just a note that I think the last statement is inaccurate: currently, a rename will not be recorded in HEAD's log. "git branch -m" will show a renaming event in the new branch's log, but the only trace of the event in HEAD's log is the deletion entry with an empty message.
Signature · 1 line
Kyle
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Jeff KingQuoted 15 lines · Jeff King
Quoted 13 lines · Jeff King
Yes. I think the options are basically (in order of decreasing preference in my opinion):
1. Log a rename entry (same sha1, but note the rename in the free-form
text).2. Log a delete (sha1 goes to null) followed by a creation (from null back to the original sha1).
3. Log nothing at all for HEAD.
This does half of (2). If we do the second half, then I'd prefer it to (3). But if we can do (1), that is better still (IMHO).
[...]
Quoted 8 lines
Quoted 4 lines
I'm actually confused about which bit of code is updating HEAD. I do not see it either in
files_rename_ref()or in the caller. Yet it clearly happens. But that is the code that would know enough to do (1) or the second half of (2) above.Ah, I found it. It's in
replace_each_worktree_head_symref()these days, which does not bother to pass a log message.
Quoted 8 lines
So I think the second half of (2) is probably something like the patch below.
Thinking on it more, we probably _do_ want two entries. Because the operations are not atomic, it's possible that we may end up in a half-way state after the first entry is written. And when debugging such a case, I'd much rather see the first half of the operation logged than nothing at all.
OK, I'll have a go at replacing patch 3 with this approach.
Signature · 1 line
Kyle
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Kyle MeyerQuoted 12 lines · Kyle Meyer
Junio C Hamano <[email protected]> writes:
[...]
Quoted 7 lines
Do we even want these "internal"
delete_ref()invocations to be logged in HEAD's reflog? I understand that this is inside the implementation of renaming an old ref to a new ref, and reflog message given todelete_ref()would matter only if the HEAD happens to be pointing at old ref---but then HEAD will be repointed to the new ref by somebody else [*1*] that called this function to rename old to new and it _will_ log it.
Quoted 5 lines
I know the discussion has developed further, but just a note that I think the last statement is inaccurate: currently, a rename will not be recorded in HEAD's log. "git branch -m" will show a renaming event in the new branch's log, but the only trace of the event in HEAD's log is the deletion entry with an empty message.
Right. I assumed Junio was talking about the hypothetical behavior we'd want.
Your response did make me think of one other option: if we updated HEAD _before_ writing the new ref, then it would automatically get the "create" half of the rename. IOW, if the rename process were:
1. delete old; this writes a reflog to HEAD, per the magic HEAD-detection in commit_ref_update().
2. update HEAD to point to new
3. re-create new; this uses the same magic to write a HEAD reflog.
That's probably a crazy path to go, though. Right now steps (1) and (3) happen in a low-level function, and step (2) happens outside of it completely. Arguably it would be good to change that, but I think we want to think of (1) and (3) as an atomic operation. Putting more things which might fail between them is a bad idea.
So I think your existing patches (modulo the review comments), plus the "something like this" that I sent on top are probably a better end state.
-Peff
Re: [PATCH 3/3] rename_ref: replace empty deletion message in HEAD's log
In reply to Kyle MeyerQuoted 12 lines · Kyle Meyer
Junio C Hamano <[email protected]> writes:
[...]
Quoted 7 lines
Do we even want these "internal"
delete_ref()invocations to be logged in HEAD's reflog? I understand that this is inside the implementation of renaming an old ref to a new ref, and reflog message given todelete_ref()would matter only if the HEAD happens to be pointing at old ref---but then HEAD will be repointed to the new ref by somebody else [*1*] that called this function to rename old to new and it _will_ log it.
Quoted 5 lines
I know the discussion has developed further, but just a note that I think the last statement is inaccurate: currently, a rename will not be recorded in HEAD's log. "git branch -m" will show a renaming event in the new branch's log, but the only trace of the event in HEAD's log is the deletion entry with an empty message.
Yes, I think Peff diagnosed it and suggested a fix.
v2 cover delete_ref: support reflog messages
In reply to Kyle MeyerThanks to Junio and Peff for their feedback on v1.
https://public-inbox.org/git/[email protected]/T/#u
Changes from v1:
* "msg" is now positioned as the first argument to delete_ref() to match update_ref().
* A "delete by head" test case has been added for the update-ref change.
* The added tests no longer send grep's output to /dev/null.
* Renaming the current branch is represented by two entries in HEAD's log, both which reuse the log message passed to rename_ref().
[email protected], [email protected]
* Corrected a few places in the commit messages where "delete_refs" was written instead of "delete_ref".
Kyle Meyer (4):
delete_ref: accept a reflog message argument
update-ref: pass reflog message to delete_ref()
rename_ref: replace empty message in HEAD's log
branch: record creation of renamed branch in HEAD's log
branch.c | 5 +++-- branch.h | 3 ++- builtin/am.c | 4 ++-- builtin/branch.c | 7 ++++--- builtin/notes.c | 4 ++-- builtin/remote.c | 4 ++-- builtin/replace.c | 2 +- builtin/reset.c | 2 +- builtin/symbolic-ref.c | 2 +- builtin/tag.c | 2 +- builtin/update-ref.c | 2 +- fast-import.c | 2 +- refs.c | 6 +++--- refs.h | 7 ++++--- refs/files-backend.c | 10 +++++----- t/t1400-update-ref.sh | 18 ++++++++++++++++++ t/t3200-branch.sh | 6 ++++++ transport.c | 2 +- 18 files changed, 58 insertions(+), 30 deletions(-)
Signature · 1 line
2.11.1
PATCH v2 1/4 delete_ref: accept a reflog message argument
When the current branch is renamed with 'git branch -m/-M' or deleted with 'git update-ref -m<msg> -d', the event is recorded in HEAD's log with an empty message. In preparation for adding a more meaningful message to HEAD's log in these cases, update delete_ref() to take a message argument and pass it along to ref_transaction_delete(). Modify all callers to pass NULL for the new message argument; no change in behavior is intended.
Note that this is relevant for HEAD's log but not for the deleted ref's log, which is currently deleted along with the ref. Even if it were not, an entry for the deletion wouldn't be present in the deleted ref's log. files_transaction_commit() writes to the log if REF_NEEDS_COMMIT or REF_LOG_ONLY are set, but lock_ref_for_update() doesn't set REF_NEEDS_COMMIT for the deleted ref because REF_DELETING is set. In contrast, the update for HEAD has REF_LOG_ONLY set by split_head_update(), resulting in the deletion being logged.
--- builtin/am.c | 4 ++-- builtin/branch.c | 2 +- builtin/notes.c | 4 ++-- builtin/remote.c | 4 ++-- builtin/replace.c | 2 +- builtin/reset.c | 2 +- builtin/symbolic-ref.c | 2 +- builtin/tag.c | 2 +- builtin/update-ref.c | 2 +- fast-import.c | 2 +- refs.c | 6 +++--- refs.h | 4 ++-- refs/files-backend.c | 6 +++--- transport.c | 2 +- 14 files changed, 22 insertions(+), 22 deletions(-)
builtin/am.c +2 -2
@@ -1049,7 +1049,7 @@ static void am_setup(struct am_state *state, enum patch_format patch_format, }else {write_state_text (state,"abort-safety" ,"" );if (!state->rebasing) -delete_ref ("ORIG_HEAD" ,NULL ,0 ); +delete_ref (NULL ,"ORIG_HEAD" ,NULL ,0 ); }/*
@@ -2172,7 +2172,7 @@ static void am_abort(struct am_state *state) has_curr_head ? &curr_head :NULL ,0 , UPDATE_REFS_DIE_ON_ERR); else if (curr_branch) - delete_ref(curr_branch, NULL, REF_NODEREF); + delete_ref(NULL, curr_branch, NULL, REF_NODEREF);free (curr_branch );am_destroy (state);
builtin/branch.c +1 -1
@@ -251,7 +251,7 @@ static int delete_branches(int argc, const char **argv, int force, int kinds, goto next; } -if (delete_ref (name,is_null_sha1 (sha1) ?NULL : sha1, +if (delete_ref (NULL , name,is_null_sha1 (sha1) ?NULL : sha1,REF_NODEREF )) { error(remote_branch ? _("Error deleting remote-tracking branch '%s'")
builtin/notes.c +2 -2
@@ -681,9 +681,9 @@ static int merge_abort(struct notes_merge_options *o)* notes_merge_abort() to remove .git/NOTES_MERGE_WORKTREE. */ -if (delete_ref ("NOTES_MERGE_PARTIAL" ,NULL ,0 )) +if (delete_ref (NULL ,"NOTES_MERGE_PARTIAL" ,NULL ,0 )) ret +=error (_ ("failed to delete ref NOTES_MERGE_PARTIAL" )); -if (delete_ref ("NOTES_MERGE_REF" ,NULL ,REF_NODEREF )) +if (delete_ref (NULL ,"NOTES_MERGE_REF" ,NULL ,REF_NODEREF )) ret +=error (_ ("failed to delete ref NOTES_MERGE_REF" ));if (notes_merge_abort (o)) ret +=error (_ ("failed to remove 'git notes merge' worktree" ));
builtin/remote.c +2 -2
@@ -691,7 +691,7 @@ static int mv(int argc, const char **argv)read_ref_full (item->string,RESOLVE_REF_READING , oid.hash, &flag);if (!(flag &REF_ISSYMREF ))continue ; -if (delete_ref (item->string,NULL ,REF_NODEREF )) +if (delete_ref (NULL , item->string,NULL ,REF_NODEREF ))die (_ ("deleting '%s' failed" ), item->string); }for (i =0 ; i < remote_branches.nr; i++) {
@@ -1248,7 +1248,7 @@ static int set_head(int argc, const char **argv) head_name =xstrdup (states.heads.items[0 ].string);free_remote_ref_states (&states); }else if (opt_d && !opt_a && argc ==1 ) { -if (delete_ref (buf.buf,NULL ,REF_NODEREF )) +if (delete_ref (NULL , buf.buf,NULL ,REF_NODEREF )) result |=error (_ ("Could not delete %s" ), buf.buf); }else usage_with_options (builtin_remote_sethead_usage, options);
builtin/replace.c +1 -1
@@ -121,7 +121,7 @@ static int for_each_replace_name(const char **argv, each_replace_name_fn fn)static int delete_replace_ref (const char *name,const char *ref,const unsigned char *sha1) { -if (delete_ref (ref, sha1,0 )) +if (delete_ref (NULL , ref, sha1,0 ))return 1 ;printf ("Deleted replace ref '%s'\n" , name);return 0 ;
builtin/reset.c +1 -1
@@ -256,7 +256,7 @@ static int reset_refs(const char *rev, const struct object_id *oid)update_ref_oid (msg.buf,"ORIG_HEAD" , orig, old_orig,0 ,UPDATE_REFS_MSG_ON_ERR ); }else if (old_orig) -delete_ref ("ORIG_HEAD" , old_orig->hash,0 ); +delete_ref (NULL ,"ORIG_HEAD" , old_orig->hash,0 );set_reflog_message (&msg,"updating HEAD" , rev); update_ref_status =update_ref_oid (msg.buf,"HEAD" , oid, orig,0 ,UPDATE_REFS_MSG_ON_ERR );
builtin/symbolic-ref.c +1 -1
@@ -58,7 +58,7 @@ int cmd_symbolic_ref(int argc, const char **argv, const char *prefix)die ("Cannot delete %s, not a symbolic ref" , argv[0 ]);if (!strcmp (argv[0 ],"HEAD" ))die ("deleting '%s' is not allowed" , argv[0 ]); -return delete_ref (argv[0 ],NULL ,REF_NODEREF ); +return delete_ref (NULL , argv[0 ],NULL ,REF_NODEREF ); }switch (argc) {
builtin/tag.c +1 -1
@@ -97,7 +97,7 @@ static int for_each_tag_name(const char **argv, each_tag_name_fn fn,static int delete_tag (const char *name,const char *ref,const unsigned char *sha1,const void *cb_data) { -if (delete_ref (ref, sha1,0 )) +if (delete_ref (NULL , ref, sha1,0 ))return 1 ;printf (_ ("Deleted tag '%s' (was %s)\n" ), name,find_unique_abbrev (sha1,DEFAULT_ABBREV ));return 0 ;
builtin/update-ref.c +1 -1
@@ -433,7 +433,7 @@ int cmd_update_ref(int argc, const char **argv, const char *prefix)* For purposes of backwards compatibility, we treat * NULL_SHA1 as "don't care" here: */ -return delete_ref (refname, +return delete_ref (NULL , refname, (oldval && !is_null_sha1 (oldsha1)) ? oldsha1 :NULL , flags); else
fast-import.c +1 -1
@@ -1752,7 +1752,7 @@ static int update_branch(struct branch *b)if (is_null_sha1 (b->sha1)) {if (b->delete) -delete_ref (b->name,NULL ,0 ); +delete_ref (NULL , b->name,NULL ,0 );return 0 ; }if (read_ref (b->name, old_sha1))
refs.c +3 -3
@@ -591,8 +591,8 @@ static int delete_pseudoref(const char *pseudoref, const unsigned char *old_sha1return 0 ; } -int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags) +int delete_ref (const char *msg,const char *refname, +const unsigned char *old_sha1,unsigned int flags) {struct ref_transaction *transaction;struct strbuf err =STRBUF_INIT ;
@@ -603,7 +603,7 @@ int delete_ref(const char *refname, const unsigned char *old_sha1, transaction =ref_transaction_begin (&err);if (!transaction ||ref_transaction_delete (transaction, refname, old_sha1, - flags,NULL , &err) || + flags, msg, &err) ||ref_transaction_commit (transaction, &err)) {error ("%s" , err.buf);ref_transaction_free (transaction);
refs.h +2 -2
@@ -276,8 +276,8 @@ int reflog_exists(const char *refname);* exists, regardless of its old value. It is an error for old_sha1 to * be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ -int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +int delete_ref (const char *msg,const char *refname, +const unsigned char *old_sha1,unsigned int flags);/* * Delete the specified references. If there are any problems, emit
refs/files-backend.c +3 -3
@@ -2489,7 +2489,7 @@ static int files_delete_refs(struct ref_store *ref_store,for (i =0 ; i < refnames->nr; i++) {const char *refname = refnames->items[i].string; -if (delete_ref (refname,NULL , flags)) +if (delete_ref (NULL , refname,NULL , flags)) result |=error (_ ("could not remove reference %s" ), refname); }
@@ -2616,7 +2616,7 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (oldrefname, orig_sha1,REF_NODEREF )) { +if (delete_ref (NULL , oldrefname, orig_sha1,REF_NODEREF )) {error ("unable to delete old %s" , oldrefname); goto rollback; }
@@ -2630,7 +2630,7 @@ static int files_rename_ref(struct ref_store *ref_store,*/ if (!read_ref_full (newrefname,RESOLVE_REF_READING |RESOLVE_REF_NO_RECURSE , sha1,NULL ) && -delete_ref (newrefname,NULL ,REF_NODEREF )) { +delete_ref (NULL , newrefname,NULL ,REF_NODEREF )) {if (errno==EISDIR ) {struct strbuf path =STRBUF_INIT ;int result;
transport.c +1 -1
@@ -299,7 +299,7 @@ void transport_update_tracking_ref(struct remote *remote, struct ref *ref, int vif (verbose)fprintf (stderr,"updating local tracking ref '%s'\n" , rs.dst);if (ref->deletion) { -delete_ref (rs.dst,NULL ,0 ); +delete_ref (NULL , rs.dst,NULL ,0 ); }else update_ref ("update by push" , rs.dst, ref->new_oid.hash,NULL ,0 ,0 );
Signature · 1 line
2.11.1
PATCH v2 2/4 update-ref: pass reflog message to delete_ref()
In reply to Kyle MeyerNow that delete_ref() accepts a reflog message, pass the user-provided message to delete_ref() rather than silently dropping it.
--- builtin/update-ref.c | 2 +- t/t1400-update-ref.sh | 18 ++++++++++++++++++ 2 files changed, 19 insertions(+), 1 deletion(-)
builtin/update-ref.c +1 -1
@@ -433,7 +433,7 @@ int cmd_update_ref(int argc, const char **argv, const char *prefix)* For purposes of backwards compatibility, we treat * NULL_SHA1 as "don't care" here: */ -return delete_ref (NULL , refname, +return delete_ref (msg, refname, (oldval && !is_null_sha1 (oldsha1)) ? oldsha1 :NULL , flags); else
t/t1400-update-ref.sh +18 -0
@@ -85,6 +85,24 @@ test_expect_success "delete $m (by HEAD)" ' ' rm -f .git/$m +test_expect_success "deleting current branch adds message to HEAD's log" ' + git update-ref $m $A && + git symbolic-ref HEAD $m && + git update-ref -m delete-$m -d $m && + ! test -f .git/$m && + grep "delete-$m$" .git/logs/HEAD +' +rm -f .git/$m + +test_expect_success "deleting by HEAD adds message to HEAD's log" ' + git update-ref $m $A && + git symbolic-ref HEAD $m && + git update-ref -m delete-by-head -d HEAD && + ! test -f .git/$m && + grep "delete-by-head$" .git/logs/HEAD +' +rm -f .git/$m + test_expect_success 'update-ref does not create reflogs by default' ' test_when_finished "git update-ref -d $outside" && git update-ref $outside $A &&
Signature · 1 line
2.11.1
PATCH v2 4/4 branch: record creation of renamed branch in HEAD's log
In reply to Kyle MeyerRenaming the current branch adds an event to the current branch's log and to HEAD's log. However, the logged entries differ. The entry in the branch's log represents the entire renaming operation (the old and new hash are identical), whereas the entry in HEAD's log represents the deletion only (the new sha1 is null).
Extend replace_each_worktree_head_symref(), whose only caller is branch_rename(), to take a reflog message argument. This allows the creation of the new ref to be recorded in HEAD's log. As a result, the renaming event is represented by two entries (a deletion and a creation entry) in HEAD's log.
It's a bit unfortunate that the branch's log and HEAD's log now represent the renaming event in different ways. Given that the renaming operation is not atomic, the two-entry form is a more accurate representation of the operation and is more useful for debugging purposes if a failure occurs between the deletion and creation events. It would make sense to move the branch's log to the two-entry form, but this would involve changes to how the rename is carried out and to how the update flags and reflogs are processed for deletions, so it may not be worth the effort.
--- branch.c | 5 +++-- branch.h | 3 ++- builtin/branch.c | 5 +++-- refs.h | 3 ++- refs/files-backend.c | 4 ++-- t/t3200-branch.sh | 5 +++-- 6 files changed, 15 insertions(+), 10 deletions(-)
branch.c +3 -2
@@ -345,7 +345,8 @@ void die_if_checked_out(const char *branch, int ignore_current_worktree) branch, wt->path); } -int replace_each_worktree_head_symref (const char *oldref,const char *newref) +int replace_each_worktree_head_symref (const char *oldref,const char *newref, +const char *logmsg) {int ret =0 ;struct worktree **worktrees =get_worktrees (0 );
@@ -358,7 +359,7 @@ int replace_each_worktree_head_symref(const char *oldref, const char *newref)continue ;if (set_worktree_head_symref (get_worktree_git_dir (worktrees[i]), - newref)) { + newref, logmsg)) { ret =-1 ;error (_ ("HEAD of working tree %s is not updated" ), worktrees[i]->path);
branch.h +2 -1
@@ -71,6 +71,7 @@ extern void die_if_checked_out(const char *branch, int ignore_current_worktree);* This will be used when renaming a branch. Returns 0 if successful, non-zero * otherwise. */ -extern int replace_each_worktree_head_symref (const char *oldref,const char *newref); +extern int replace_each_worktree_head_symref (const char *oldref,const char *newref, +const char *logmsg);#endif
builtin/branch.c +3 -2
@@ -579,14 +579,15 @@ static void rename_branch(const char *oldname, const char *newname, int force)if (rename_ref (oldref.buf, newref.buf, logmsg.buf))die (_ ("Branch rename failed" )); -strbuf_release (&logmsg);if (recovery)warning (_ ("Renamed a misnamed branch '%s' away" ), oldref.buf +11 ); -if (replace_each_worktree_head_symref (oldref.buf, newref.buf)) +if (replace_each_worktree_head_symref (oldref.buf, newref.buf, logmsg.buf))die (_ ("Branch renamed to %s, but HEAD is not updated!" ), newname); +strbuf_release (&logmsg); +strbuf_addf (&oldsection,"branch.%s" , oldref.buf +11 );strbuf_release (&oldref);strbuf_addf (&newsection,"branch.%s" , newref.buf +11 );
refs.h +2 -1
@@ -334,7 +334,8 @@ int create_symref(const char *refname, const char *target, const char *logmsg);* $GIT_DIR points to. * Return 0 if successful, non-zero otherwise. * */ -int set_worktree_head_symref (const char *gitdir,const char *target); +int set_worktree_head_symref (const char *gitdir,const char *target, +const char *logmsg);enum action_on_err {UPDATE_REFS_MSG_ON_ERR ,
refs/files-backend.c +2 -2
@@ -3055,7 +3055,7 @@ static int files_create_symref(struct ref_store *ref_store,return ret; } -int set_worktree_head_symref (const char *gitdir,const char *target) +int set_worktree_head_symref (const char *gitdir,const char *target,const char *logmsg) {static struct lock_file head_lock;struct ref_lock *lock;
@@ -3083,7 +3083,7 @@ int set_worktree_head_symref(const char *gitdir, const char *target) lock->lk = &head_lock; lock->ref_name =xstrdup (head_rel); - ret =create_symref_locked (lock, head_rel, target,NULL ); + ret =create_symref_locked (lock, head_rel, target, logmsg);unlock_ref (lock);/* will free lock */ strbuf_release (&head_path);
t/t3200-branch.sh +3 -2
@@ -139,9 +139,10 @@ test_expect_success 'git branch -M baz bam should succeed when baz is checked ou test $(git rev-parse --abbrev-ref HEAD) = bam ' -test_expect_success 'git branch -M baz bam should add entry to .git/logs/HEAD' ' +test_expect_success 'git branch -M baz bam should add entries to .git/logs/HEAD' ' msg="Branch: renamed refs/heads/baz to refs/heads/bam" && - grep " 0\{40\}.*$msg$" .git/logs/HEAD + grep " 0\{40\}.*$msg$" .git/logs/HEAD && + grep "^0\{40\}.*$msg$" .git/logs/HEAD ' test_expect_success 'git branch -M baz bam should succeed when baz is checked out as linked working tree' '
Signature · 1 line
2.11.1
PATCH v2 3/4 rename_ref: replace empty message in HEAD's log
In reply to Kyle MeyerWhen the current branch is renamed, the deletion of the old ref is recorded in HEAD's log with an empty message. Now that delete_ref() accepts a reflog message, provide a more descriptive message by passing along the log message that is given to rename_ref().
The next step will be to extend HEAD's log to also include the second part of the rename, the creation of the new branch.
--- refs/files-backend.c | 2 +- t/t3200-branch.sh | 5 +++++ 2 files changed, 6 insertions(+), 1 deletion(-)
refs/files-backend.c +1 -1
@@ -2616,7 +2616,7 @@ static int files_rename_ref(struct ref_store *ref_store,return error ("unable to move logfile logs/%s to " TMP_RENAMED_LOG ": %s" , oldrefname,strerror (errno)); -if (delete_ref (NULL , oldrefname, orig_sha1,REF_NODEREF )) { +if (delete_ref (logmsg, oldrefname, orig_sha1,REF_NODEREF )) {error ("unable to delete old %s" , oldrefname); goto rollback; }
t/t3200-branch.sh +5 -0
@@ -139,6 +139,11 @@ test_expect_success 'git branch -M baz bam should succeed when baz is checked ou test $(git rev-parse --abbrev-ref HEAD) = bam ' +test_expect_success 'git branch -M baz bam should add entry to .git/logs/HEAD' ' + msg="Branch: renamed refs/heads/baz to refs/heads/bam" && + grep " 0\{40\}.*$msg$" .git/logs/HEAD +' + test_expect_success 'git branch -M baz bam should succeed when baz is checked out as linked working tree' ' git checkout master && git worktree add -b baz bazdir &&
Signature · 1 line
2.11.1
Re: [PATCH v2 0/4] delete_ref: support reflog messages
In reply to Kyle MeyerQuoted 5 lines · Kyle Meyer
Kyle Meyer (4):
delete_ref: accept a reflog message argument
update-ref: pass reflog message todelete_ref()
rename_ref: replace empty message in HEAD's log
branch: record creation of renamed branch in HEAD's log
These look good to me. I think you did a nice job of summarizing the discussion in the commit messages.
-Peff
Re: [PATCH v2 0/4] delete_ref: support reflog messages
In reply to Kyle MeyerQuoted 5 lines · Kyle Meyer
Kyle Meyer (4):
delete_ref: accept a reflog message argument
update-ref: pass reflog message todelete_ref()
rename_ref: replace empty message in HEAD's log
branch: record creation of renamed branch in HEAD's log
These looked reasonable. I had to resolve conflicts with another topic in flight that removed set_worktree_head_symref() function while queuing, and I think I resolved it correctly, but please double check what is pushed out on 'pu'.
Thanks.
Re: [PATCH v2 0/4] delete_ref: support reflog messages
Quoted 4 lines · Junio C Hamano
These looked reasonable. I had to resolve conflicts with another topic in flight that removed
set_worktree_head_symref()function while queuing, and I think I resolved it correctly, but please double check what is pushed out on 'pu'.
The merge looks right to me, thanks. Thanks also for touching up the tab/space issue in t3200-branch.sh.
Signature · 1 line
Kyle
Re: [PATCH v2 1/4] delete_ref: accept a reflog message argument
In reply to Kyle Meyerrefs.h int reflog_exists(const char *refname); · line 280
Quoted 12 lines · Kyle Meyer
refs.h +2
-2@@ -276,8 +276,8 @@ int reflog_exists(const char *refname);* exists, regardless of its old value. It is an error for old_sha1 to * be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ -int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +int delete_ref (const char *msg,const char *refname, +const unsigned char *old_sha1,unsigned int flags);
Is it just me who thinks it's weird that msg comes in front here? The key identity, refname, should be the first argument imo. You'll probably want to update the comment block above if msg can be NULL. We have _very_ good documentation in this file, let's keep it uptodate.
Signature · 1 line
Duy
Re: [PATCH v2 1/4] delete_ref: accept a reflog message argument
Quoted 15 lines · Duy Nguyen
refs.h int reflog_exists(const char *refname); · line 280
Quoted 12 lines · Kyle Meyer
refs.h +2
-2@@ -276,8 +276,8 @@ int reflog_exists(const char *refname);* exists, regardless of its old value. It is an error for old_sha1 to * be NULL_SHA1. flags is passed through to ref_transaction_delete(). */ -int delete_ref (const char *refname,const unsigned char *old_sha1, -unsigned int flags); +int delete_ref (const char *msg,const char *refname, +const unsigned char *old_sha1,unsigned int flags);Is it just me who thinks it's weird that msg comes in front here?
You and anybody who didn't read what was discussed earlier, methinks ;-) cf. <[email protected]>
Quoted 3 lines
... You'll
probably want to update the comment block above if msg can be NULL.
Good suggestion.
Thanks for taking a look at this topic. IIRC a recent update to one of your topics introduced a new conflicts with this one.
Re: [PATCH v2 1/4] delete_ref: accept a reflog message argument
In reply to Duy NguyenQuoted 3 lines · Duy Nguyen
You'll probably want to update the comment block above if msg can be NULL. We have _very_ good documentation in this file, let's keep it uptodate.
Looking at how other functions in refs.h document their "msg" or "logmsg" parameter, none seem to mention it explicitly. update_ref refers to ref_transaction_update, and ref_transaction_{update,create,delete,verify} refer to ref.h's "Reference transaction updates" comment.
delete_ref's docstring already mentions that "flag" is passed to ref_transaction_delete, so perhaps it should mention "msg" here as well.
-- >8 --
Subject: [PATCH] delete_ref: mention "msg" parameter in docstring
delete_ref() was recently extended to take a "msg" argument.
--- refs.h | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
refs.h +2 -1
@@ -274,7 +274,8 @@ int reflog_exists(const char *refname);* verify that the current value of the reference is old_sha1 before * deleting it. If old_sha1 is NULL, delete the reference if it * exists, regardless of its old value. It is an error for old_sha1 to -* be NULL_SHA1. flags is passed through to ref_transaction_delete(). +* be NULL_SHA1. msg and flags are passed through to +* ref_transaction_delete(). */ int delete_ref (const char *msg,const char *refname,const unsigned char *old_sha1,unsigned int flags);
Signature · 1 line
2.11.1