use delete_refs when deleting tags or branches
PATCH v3 1/1 use delete_refs when deleting tags or branches
From: Phil Hord <[email protected]>
'git tag -d' accepts one or more tag refs to delete, but each deletion is done by calling delete_ref on each argv. This is very slow when removing from packed refs. Use delete_refs instead so all the removals can be done inside a single transaction with a single update.
Do the same for 'git branch -d'.
Since delete_refs performs all the packed-refs delete operations inside a single transaction, if any of the deletes fail then all them will be skipped. In practice, none of them should fail since we verify the hash of each one before calling delete_refs, but some network error or odd permissions problem could have different results after this change.
Also, since the file-backed deletions are not performed in the same transaction, those could succeed even when the packed-refs transaction fails.
After deleting branches, remove the branch config only if the branch ref was removed and was not subsequently added back in.
A manual test deleting 24,000 tags took about 30 minutes using delete_ref. It takes about 5 seconds using delete_refs.
---
This version translates a nonzero return code from delete_refs into an error return value of 1. It also has style cleanups suggested from v2.
builtin/branch.c | 47 ++++++++++++++++++++++++++++------------------- builtin/tag.c | 44 ++++++++++++++++++++++++++++++++++---------- 2 files changed, 62 insertions(+), 29 deletions(-)
branch.c +28 -19
@@ -202,6 +202,9 @@ static int delete_branches(int argc, const char **argv, int force, int kinds,int remote_branch =0 ;struct strbuf bname =STRBUF_INIT ;unsigned allowed_interpret; +struct string_list refs_to_delete =STRING_LIST_INIT_DUP ; +struct string_list_item *item; +int branch_name_pos;switch (kinds) {case FILTER_REFS_REMOTES :
@@ -219,6 +222,7 @@ static int delete_branches(int argc, const char **argv, int force, int kinds,default :die (_ ("cannot use -a with -d" )); } + branch_name_pos =strcspn (fmt,"%" );if (!force) { head_rev =lookup_commit_reference (the_repository, &head_oid);
@@ -265,30 +269,35 @@ static int delete_branches(int argc, const char **argv, int force, int kinds, goto next; } -if (delete_ref (NULL , name,is_null_oid (&oid) ?NULL : &oid, -REF_NO_DEREF )) { -error (remote_branch - ?_ ("Error deleting remote-tracking branch '%s'" ) - :_ ("Error deleting branch '%s'" ), - bname.buf); - ret =1 ; - goto next; - } -if (!quiet) { -printf (remote_branch - ?_ ("Deleted remote-tracking branch %s (was %s).\n" ) - :_ ("Deleted branch %s (was %s).\n" ), - bname.buf, - (flags &REF_ISBROKEN ) ?"broken" - : (flags &REF_ISSYMREF ) ? target - :find_unique_abbrev (&oid,DEFAULT_ABBREV )); - } -delete_branch_config (bname.buf); + item =string_list_append (&refs_to_delete, name); + item->util =xstrdup ((flags &REF_ISBROKEN ) ?"broken" + : (flags &REF_ISSYMREF ) ? target + :find_unique_abbrev (&oid,DEFAULT_ABBREV )); next:free (target); } +if (delete_refs (NULL , &refs_to_delete,REF_NO_DEREF )) + ret =1 ; + +for_each_string_list_item (item, &refs_to_delete) { +char *describe_ref = item->util; +char *name = item->string; +if (!ref_exists (name)) { +char *refname = name + branch_name_pos; +if (!quiet) +printf (remote_branch + ?_ ("Deleted remote-tracking branch %s (was %s).\n" ) + :_ ("Deleted branch %s (was %s).\n" ), + name + branch_name_pos, describe_ref); + +delete_branch_config (refname); + } +free (describe_ref); + } +string_list_clear (&refs_to_delete,0 ); +free (name);strbuf_release (&bname);
tag.c +34 -10
69 lines not on this page · raw message
Signature · 1 line
2.30.0.281.g914876b2ce
branch.c
Quoted 4 lines · Phil Hord
branch.c +0
-0
You wasted 15 minutes of my life by choosing to deviate the list norm of sending what "git apply -p1" (default) would accept. I am fairly trusting type and did not suspect anybody do such an evil thing. Why?
Quoted 12 lines · Junio C Hamano
branch.c
Quoted 4 lines · Phil Hord
branch.c +0
-0You wasted 15 minutes of my life by choosing to deviate the list norm of sending what "git apply -p1" (default) would accept. I am fairly trusting type and did not suspect anybody do such an evil thing. Why?
Oof. Sorry. I forgot I have diff.noprefix=true in my local config. It is a huge timesaver for me when looking at diffs on a console since I can quickly highlight the filename with a mouse to paste into an editor.
Sometimes it bites me, though. Usually I notice in the diff, but this one I was sending with format-patch / send-email.
I guess I'll turn that off in git.git so I don't misfire at you again someday.
Phil
Quoted 9 lines · Phil Hord
Oof. Sorry. I forgot I have diff.noprefix=true in my local config. It is a huge timesaver for me when looking at diffs on a console since I can quickly highlight the filename with a mouse to paste into an editor.
Sometimes it bites me, though. Usually I notice in the diff, but this one I was sending with format-patch / send-email.
I guess I'll turn that off in git.git so I don't misfire at you again someday.
I think per-repository configuration might be sufficient for this particular case (after all, it is project's preference), I wonder if a more command-specific variant of diff.noprefix so that "log -p" and "format-patch" can be configured separately would make sense, something like...
[diff]
noprefix = true
[diff "format-patch"]
noprefix = false
[diff "show"]
noprefix = false
Quoted 19 lines · Junio C Hamano
Quoted 9 lines · Phil Hord
Oof. Sorry. I forgot I have diff.noprefix=true in my local config. It is a huge timesaver for me when looking at diffs on a console since I can quickly highlight the filename with a mouse to paste into an editor.
Sometimes it bites me, though. Usually I notice in the diff, but this one I was sending with format-patch / send-email.
I guess I'll turn that off in git.git so I don't misfire at you again someday.
I think per-repository configuration might be sufficient for this particular case (after all, it is project's preference), I wonder if a more command-specific variant of diff.noprefix so that "log -p" and "format-patch" can be configured separately would make sense, something like...
Quoted 6 lines
[diff]
noprefix = true
[diff "format-patch"]
noprefix = false
[diff "show"]
noprefix = false
That seems reasonable. I was trying to think of something clever like a setting like "auto" that means "noprefix when output is to a tty". But I still sometimes send a patch to a coworker that I copied from my console and I then have to remember to add back in the prefixes. So there is no perfect solution from Git, I think. The correct solution is to teach my console to skip the prefixes when I double-click the filename; or to add symlinks at a and b in my project; or something else. But these are all more painful than noprefix = true, so far.
Fwiw - I know this issue has been discussed on the list before; there are others who feel this itch.
Having config diff.<command>.noprefix seems reasonable as a fix for format-patch. At first glance it seems like this could get confused with diff.<driver>.*, but I suppose those settings are all specific to a driver section, so it would be easy enough to keep them separate.
Quoted 5 lines · Phil Hord
After deleting branches, remove the branch config only if the branch ref was removed and was not subsequently added back in.
A manual test deleting 24,000 tags took about 30 minutes using
delete_ref. It takes about 5 seconds usingdelete_refs.
Nicely done. Queued and pushed out.