| Message ID | 95ffa62df6ce394249a8ddabb84fb2b517825fe3.1740901525.git.gitgitgadget@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | completion: fix bugs with slashes in remote names | expand |
Hi David On 02/03/2025 07:45, David Mandelberg via GitGitGadget wrote: > From: David Mandelberg <david@mandelberg.org> > > Previously, some calls to for-each-ref passed fixed numbers of path > components to strip from refs, assuming that remote names had no slashes > in them. This made completions like: > > git push github/dseomn :com<Tab> > > Result in: > > git push github/dseomn :dseomn/completion-remote-slash > > With this patch, it instead results in: > > git push github/dseomn :completion-remote-slash This sounds like a useful improvement and I like the idea, but I think running "git for-each-ref" once for each remote is not going to scale very well for people who have a lot of remotes. I think it would be better to try and strip "refs/remote/$remote/" outside of "git for-each-ref". I've not tested it but I think something like local sed_cmd= local remote # ref names and therefore remote names cannot contain '*?[]^' so we # only need to escape '.$/'. Using 'sort -r' means that if there is a # remote called "github" and another called "github/alice" we will try # and strip "github/alice" first. for remote in $(__git_remotes | sort -r) do remote="${remote//./\\./}" remote="${remote//\$/\\\$/}" remote="${remote//\//\\\//}" sed_cmd="${sed_cmd} -e s/^refs\/remotes\/$remote\/// -e t" done __git for-each-ref --format="$fer_pfx$sfx" \ ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | sed $sed_cmd | sort -u should work and means we're only forking three extra processes no matter how many remotes the user has. I'm not sure if it changes the output order when GIT_COMPLETION_IGNORE_CASE is set though. Best Wishes Phillip > In addition, the change to __git_dwim_remote_heads() restricts it to > only list remote branches, not HEAD. I think that actually improves the > completion where it's used though, since HEAD doesn't seem to be valid > anyway: > > $ git switch HEAD > fatal: a branch is expected, got 'HEAD' > > Signed-off-by: David Mandelberg <david@mandelberg.org> > --- > contrib/completion/git-completion.bash | 17 ++- > t/t9902-completion.sh | 180 ++++++++++++++++++++++--- > 2 files changed, 169 insertions(+), 28 deletions(-) > > diff --git a/contrib/completion/git-completion.bash b/contrib/completion/git-completion.bash > index 17c044f7d60..5f4407be37f 100644 > --- a/contrib/completion/git-completion.bash > +++ b/contrib/completion/git-completion.bash > @@ -807,11 +807,15 @@ __git_dwim_remote_heads () > # employ the heuristic used by git checkout and git switch > # Try to find a remote branch that cur_es the completion word > # but only output if the branch name is unique > - __git for-each-ref --format="$fer_pfx%(refname:strip=3)$sfx" \ > - --sort="refname:strip=3" \ > - ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ > - "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | \ > - uniq -u > + local remote > + for remote in $(__git_remotes); do > + local base="$(__git_escape_fnmatch "refs/remotes/$remote")" > + local strip="$(__git_count_path_components "$base")" > + __git for-each-ref \ > + --format="$fer_pfx%(refname:strip=$strip)$sfx" \ > + ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ > + "$base/$cur_*" "$base/$cur_*/**" > + done | sort | uniq -u > } > > # Lists refs from the local (by default) or from a remote repository. > @@ -917,7 +921,8 @@ __git_refs () > case "HEAD" in > $match*|$umatch*) echo "${pfx}HEAD$sfx" ;; > esac > - __git for-each-ref --format="$fer_pfx%(refname:strip=3)$sfx" \ > + local strip="$(__git_count_path_components "refs/remotes/$remote")" > + __git for-each-ref --format="$fer_pfx%(refname:strip=$strip)$sfx" \ > ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ > "refs/remotes/$remote/$match*" \ > "refs/remotes/$remote/$match*/**" > diff --git a/t/t9902-completion.sh b/t/t9902-completion.sh > index 0219408358c..0c41ef9d5b1 100755 > --- a/t/t9902-completion.sh > +++ b/t/t9902-completion.sh > @@ -149,7 +149,8 @@ fi > test_expect_success 'setup for __git_find_repo_path/__gitdir tests' ' > mkdir -p subdir/subsubdir && > mkdir -p non-repo && > - git init -b main otherrepo > + git init -b main otherrepo && > + git init -b main slashrepo > ' > > test_expect_success '__git_find_repo_path - from command line (through $__git_dir)' ' > @@ -681,6 +682,13 @@ test_expect_success 'setup for ref completion' ' > ) && > git remote add other "$ROOT/otherrepo/.git" && > git fetch --no-tags other && > + ( > + cd slashrepo && > + git commit --allow-empty -m initial && > + git branch -m main branch/with/slash > + ) && > + git remote add remote/with/slash "$ROOT/slashrepo/.git" && > + git fetch --no-tags remote/with/slash && > rm -f .git/FETCH_HEAD && > git init thirdrepo > ' > @@ -693,6 +701,8 @@ test_expect_success '__git_refs - simple' ' > other/HEAD > other/branch-in-other > other/main-in-other > + remote/with/slash/HEAD > + remote/with/slash/branch/with/slash > matching-tag > EOF > ( > @@ -709,6 +719,8 @@ test_expect_success '__git_refs - full refs' ' > refs/remotes/other/HEAD > refs/remotes/other/branch-in-other > refs/remotes/other/main-in-other > + refs/remotes/remote/with/slash/HEAD > + refs/remotes/remote/with/slash/branch/with/slash > refs/tags/matching-tag > EOF > ( > @@ -774,6 +786,19 @@ test_expect_success '__git_refs - configured remote' ' > test_cmp expected "$actual" > ' > > +test_expect_success '__git_refs - configured remote - with slash' ' > + cat >expected <<-EOF && > + HEAD > + HEAD > + branch/with/slash > + EOF > + ( > + cur= && > + __git_refs remote/with/slash >"$actual" > + ) && > + test_cmp expected "$actual" > +' > + > test_expect_success '__git_refs - configured remote - full refs' ' > cat >expected <<-EOF && > HEAD > @@ -916,17 +941,19 @@ test_expect_success '__git_refs - unique remote branches for git checkout DWIMer > other/ambiguous > other/branch-in-other > other/main-in-other > - remote/ambiguous > - remote/branch-in-remote > + remote/with/slash/HEAD > + remote/with/slash/ambiguous > + remote/with/slash/branch-in-remote > + remote/with/slash/branch/with/slash > matching-tag > - HEAD > branch-in-other > branch-in-remote > + branch/with/slash > main-in-other > EOF > for remote_ref in refs/remotes/other/ambiguous \ > - refs/remotes/remote/ambiguous \ > - refs/remotes/remote/branch-in-remote > + refs/remotes/remote/with/slash/ambiguous \ > + refs/remotes/remote/with/slash/branch-in-remote > do > git update-ref $remote_ref main && > test_when_finished "git update-ref -d $remote_ref" || return 1 > @@ -946,6 +973,8 @@ test_expect_success '__git_refs - after --opt=' ' > other/HEAD > other/branch-in-other > other/main-in-other > + remote/with/slash/HEAD > + remote/with/slash/branch/with/slash > matching-tag > EOF > ( > @@ -962,6 +991,8 @@ test_expect_success '__git_refs - after --opt= - full refs' ' > refs/remotes/other/HEAD > refs/remotes/other/branch-in-other > refs/remotes/other/main-in-other > + refs/remotes/remote/with/slash/HEAD > + refs/remotes/remote/with/slash/branch/with/slash > refs/tags/matching-tag > EOF > ( > @@ -979,6 +1010,8 @@ test_expect_success '__git refs - excluding refs' ' > ^other/HEAD > ^other/branch-in-other > ^other/main-in-other > + ^remote/with/slash/HEAD > + ^remote/with/slash/branch/with/slash > ^matching-tag > EOF > ( > @@ -995,6 +1028,8 @@ test_expect_success '__git refs - excluding full refs' ' > ^refs/remotes/other/HEAD > ^refs/remotes/other/branch-in-other > ^refs/remotes/other/main-in-other > + ^refs/remotes/remote/with/slash/HEAD > + ^refs/remotes/remote/with/slash/branch/with/slash > ^refs/tags/matching-tag > EOF > ( > @@ -1022,6 +1057,8 @@ test_expect_success '__git_refs - do not filter refs unless told so' ' > other/branch-in-other > other/main-in-other > other/matching/branch-in-other > + remote/with/slash/HEAD > + remote/with/slash/branch/with/slash > matching-tag > matching/tag > EOF > @@ -1142,6 +1179,8 @@ test_expect_success '__git_complete_refs - simple' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > matching-tag Z > EOF > ( > @@ -1180,6 +1219,20 @@ test_expect_success '__git_complete_refs - remote' ' > test_cmp expected out > ' > > +test_expect_success '__git_complete_refs - remote - with slash' ' > + sed -e "s/Z$//" >expected <<-EOF && > + HEAD Z > + HEAD Z > + branch/with/slash Z > + EOF > + ( > + cur= && > + __git_complete_refs --remote=remote/with/slash && > + print_comp > + ) && > + test_cmp expected out > +' > + > test_expect_success '__git_complete_refs - track' ' > sed -e "s/Z$//" >expected <<-EOF && > HEAD Z > @@ -1188,9 +1241,11 @@ test_expect_success '__git_complete_refs - track' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > matching-tag Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main-in-other Z > EOF > ( > @@ -1235,6 +1290,8 @@ test_expect_success '__git_complete_refs - suffix' ' > other/HEAD. > other/branch-in-other. > other/main-in-other. > + remote/with/slash/HEAD. > + remote/with/slash/branch/with/slash. > matching-tag. > EOF > ( > @@ -1260,6 +1317,20 @@ test_expect_success '__git_complete_fetch_refspecs - simple' ' > test_cmp expected out > ' > > +test_expect_success '__git_complete_fetch_refspecs - with slash' ' > + sed -e "s/Z$//" >expected <<-EOF && > + HEAD:HEAD Z > + HEAD:HEAD Z > + branch/with/slash:branch/with/slash Z > + EOF > + ( > + cur= && > + __git_complete_fetch_refspecs remote/with/slash && > + print_comp > + ) && > + test_cmp expected out > +' > + > test_expect_success '__git_complete_fetch_refspecs - matching' ' > sed -e "s/Z$//" >expected <<-EOF && > branch-in-other:branch-in-other Z > @@ -1340,8 +1411,8 @@ test_expect_success '__git_complete_worktree_paths with -C' ' > > test_expect_success 'git switch - with no options, complete local branches and unique remote branch names for DWIM logic' ' > test_completion "git switch " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1487,8 +1558,8 @@ test_expect_success 'git-bisect - existing view subcommand is recognized and ena > test_expect_success 'git checkout - completes refs and unique remote branches for DWIM' ' > test_completion "git checkout " <<-\EOF > HEAD Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1496,6 +1567,8 @@ test_expect_success 'git checkout - completes refs and unique remote branches fo > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1515,8 +1588,8 @@ test_expect_success 'git switch - with GIT_COMPLETION_CHECKOUT_NO_GUESS=1, compl > > test_expect_success 'git switch - --guess overrides GIT_COMPLETION_CHECKOUT_NO_GUESS=1, complete local branches and unique remote names for DWIM logic' ' > GIT_COMPLETION_CHECKOUT_NO_GUESS=1 test_completion "git switch --guess " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1525,8 +1598,8 @@ test_expect_success 'git switch - --guess overrides GIT_COMPLETION_CHECKOUT_NO_G > > test_expect_success 'git switch - a later --guess overrides previous --no-guess, complete local and remote unique branches for DWIM' ' > test_completion "git switch --no-guess --guess " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1549,14 +1622,16 @@ test_expect_success 'git checkout - with GIT_COMPLETION_NO_GUESS=1 only complete > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > test_expect_success 'git checkout - --guess overrides GIT_COMPLETION_NO_GUESS=1, complete refs and unique remote branches for DWIM' ' > GIT_COMPLETION_CHECKOUT_NO_GUESS=1 test_completion "git checkout --guess " <<-\EOF > HEAD Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1564,6 +1639,8 @@ test_expect_success 'git checkout - --guess overrides GIT_COMPLETION_NO_GUESS=1, > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1576,14 +1653,16 @@ test_expect_success 'git checkout - with --no-guess, only completes refs' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > test_expect_success 'git checkout - a later --guess overrides previous --no-guess, complete refs and unique remote branches for DWIM' ' > test_completion "git checkout --no-guess --guess " <<-\EOF > HEAD Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1591,6 +1670,8 @@ test_expect_success 'git checkout - a later --guess overrides previous --no-gues > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1603,6 +1684,8 @@ test_expect_success 'git checkout - a later --no-guess overrides previous --gues > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1616,6 +1699,8 @@ test_expect_success 'git checkout - with checkout.guess = false, only completes > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1623,8 +1708,8 @@ test_expect_success 'git checkout - with checkout.guess = true, completes refs a > test_config checkout.guess true && > test_completion "git checkout " <<-\EOF > HEAD Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1632,6 +1717,8 @@ test_expect_success 'git checkout - with checkout.guess = true, completes refs a > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1639,8 +1726,8 @@ test_expect_success 'git checkout - a later --guess overrides previous checkout. > test_config checkout.guess false && > test_completion "git checkout --guess " <<-\EOF > HEAD Z > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -1648,6 +1735,8 @@ test_expect_success 'git checkout - a later --guess overrides previous checkout. > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1661,6 +1750,8 @@ test_expect_success 'git checkout - a later --no-guess overrides previous checko > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1673,6 +1764,8 @@ test_expect_success 'git switch - with --detach, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1685,6 +1778,8 @@ test_expect_success 'git checkout - with --detach, complete only references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1857,6 +1952,8 @@ test_expect_success 'git switch - with -d, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1869,6 +1966,8 @@ test_expect_success 'git checkout - with -d, complete only references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1877,11 +1976,15 @@ test_expect_success 'git switch - with --track, complete only remote branches' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > test_completion "git switch -t " <<-\EOF > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1890,11 +1993,15 @@ test_expect_success 'git checkout - with --track, complete only remote branches' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > test_completion "git checkout -t " <<-\EOF > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1914,6 +2021,8 @@ test_expect_success 'git checkout - with --no-track, complete only local referen > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1926,6 +2035,8 @@ test_expect_success 'git switch - with -c, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1938,6 +2049,8 @@ test_expect_success 'git switch - with -C, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1950,6 +2063,8 @@ test_expect_success 'git switch - with -c and --track, complete all references' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1962,6 +2077,8 @@ test_expect_success 'git switch - with -C and --track, complete all references' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1974,6 +2091,8 @@ test_expect_success 'git switch - with -c and --no-track, complete all reference > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1986,6 +2105,8 @@ test_expect_success 'git switch - with -C and --no-track, complete all reference > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -1998,6 +2119,8 @@ test_expect_success 'git checkout - with -b, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2010,6 +2133,8 @@ test_expect_success 'git checkout - with -B, complete all references' ' > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2022,6 +2147,8 @@ test_expect_success 'git checkout - with -b and --track, complete all references > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2034,6 +2161,8 @@ test_expect_success 'git checkout - with -B and --track, complete all references > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2046,6 +2175,8 @@ test_expect_success 'git checkout - with -b and --no-track, complete all referen > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2058,13 +2189,15 @@ test_expect_success 'git checkout - with -B and --no-track, complete all referen > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > test_expect_success 'git switch - for -c, complete local branches and unique remote branches' ' > test_completion "git switch -c " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2073,8 +2206,8 @@ test_expect_success 'git switch - for -c, complete local branches and unique rem > > test_expect_success 'git switch - for -C, complete local branches and unique remote branches' ' > test_completion "git switch -C " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2111,8 +2244,8 @@ test_expect_success 'git switch - for -C with --no-track, complete local branche > > test_expect_success 'git checkout - for -b, complete local branches and unique remote branches' ' > test_completion "git checkout -b " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2121,8 +2254,8 @@ test_expect_success 'git checkout - for -b, complete local branches and unique r > > test_expect_success 'git checkout - for -B, complete local branches and unique remote branches' ' > test_completion "git checkout -B " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2159,8 +2292,8 @@ test_expect_success 'git checkout - for -B with --no-track, complete local branc > > test_expect_success 'git switch - with --orphan completes local branch names and unique remote branch names' ' > test_completion "git switch --orphan " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2175,8 +2308,8 @@ test_expect_success 'git switch - --orphan with branch already provided complete > > test_expect_success 'git checkout - with --orphan completes local branch names and unique remote branch names' ' > test_completion "git checkout --orphan " <<-\EOF > - HEAD Z > branch-in-other Z > + branch/with/slash Z > main Z > main-in-other Z > matching-branch Z > @@ -2192,6 +2325,8 @@ test_expect_success 'git checkout - --orphan with branch already provided comple > other/HEAD Z > other/branch-in-other Z > other/main-in-other Z > + remote/with/slash/HEAD Z > + remote/with/slash/branch/with/slash Z > EOF > ' > > @@ -2206,7 +2341,8 @@ test_expect_success 'git restore completes modified files' ' > test_expect_success 'teardown after ref completion' ' > git branch -d matching-branch && > git tag -d matching-tag && > - git remote remove other > + git remote remove other && > + git remote remove remote/with/slash > ' > >
Op 2025-03-02 om 09:17 schreef Phillip Wood: > Hi David > > On 02/03/2025 07:45, David Mandelberg via GitGitGadget wrote: >> From: David Mandelberg <david@mandelberg.org> >> >> Previously, some calls to for-each-ref passed fixed numbers of path >> components to strip from refs, assuming that remote names had no slashes >> in them. This made completions like: >> >> git push github/dseomn :com<Tab> >> >> Result in: >> >> git push github/dseomn :dseomn/completion-remote-slash >> >> With this patch, it instead results in: >> >> git push github/dseomn :completion-remote-slash > > This sounds like a useful improvement and I like the idea, but I think > running "git for-each-ref" once for each remote is not going to scale > very well for people who have a lot of remotes. I think it would be > better to try and strip "refs/remote/$remote/" outside of "git for-each- > ref". I've not tested it but I think something like Good point, I hadn't thought of that. Do you have a rough estimate of what "a lot of remotes" is? 100ish, maybe? I'd like to do some testing to get actual performance numbers before trying to optimize this, because I think the optimization has some drawbacks, see below. If optimization is needed, another approach is to parallelize the forks: { local fer_pids= for ... do __git for-each-ref ... & fer_pids="$fer_pids $!" done test -z "$fer_pids" || wait $fer_pids } | sort | uniq -u That might cause spikes in cpu/memory/disk usage that aren't ideal though. > local sed_cmd= > local remote > # ref names and therefore remote names cannot contain '*?[]^' so we > # only need to escape '.$/'. Using 'sort -r' means that if there is a > # remote called "github" and another called "github/alice" we will try > # and strip "github/alice" first. > for remote in $(__git_remotes | sort -r) > do > remote="${remote//./\\./}" > remote="${remote//\$/\\\$/}" > remote="${remote//\//\\\//}" Just FYI since it took me hours to figure this out myself: I think this would break tests on macos because of an old version of bash that handles backslashes weirdly. I think removing the double quotes would work around that issue, and be safe because word splitting doesn't happen in assignments. > sed_cmd="${sed_cmd} -e s/^refs\/remotes\/$remote\/// -e t" > done Mostly just a note to myself if I end up using this idea: I think a space in $remote would break this because bash would split up the arg to sed. There's probably some way to fix that with extra escaping though? > __git for-each-ref --format="$fer_pfx$sfx" \ > ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ > "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | This would search for $cur_ in the wrong place because * only matches one path component, right? Changing to ** might help, but then it would match in places more it shouldn't. > sed $sed_cmd | sort -u > > should work and means we're only forking three extra processes no matter > how many remotes the user has. I'm not sure if it changes the output > order when GIT_COMPLETION_IGNORE_CASE is set though. > > Best Wishes > > Phillip
Hi David On 02/03/2025 20:34, David Mandelberg wrote: > Op 2025-03-02 om 09:17 schreef Phillip Wood: >> On 02/03/2025 07:45, David Mandelberg via GitGitGadget wrote: >>> From: David Mandelberg <david@mandelberg.org> >>> >> This sounds like a useful improvement and I like the idea, but I think >> running "git for-each-ref" once for each remote is not going to scale >> very well for people who have a lot of remotes. I think it would be >> better to try and strip "refs/remote/$remote/" outside of "git for- >> each- ref". I've not tested it but I think something like > > Good point, I hadn't thought of that. Do you have a rough estimate of > what "a lot of remotes" is? 100ish, maybe? I'm not really sure what a "large" number looks like for people who track a lot of repositories. On windows forking is pretty slow (our shell-based test suite is incredibly slow on that platform) so I think it makes sense to try and avoid adding new processes. > I'd like to do some testing > to get actual performance numbers before trying to optimize this, > because I think the optimization has some drawbacks, see below. > > If optimization is needed, another approach is to parallelize the forks: > > { > local fer_pids= > for ... > do > __git for-each-ref ... & > fer_pids="$fer_pids $!" > done > test -z "$fer_pids" || wait $fer_pids > } | sort | uniq -u > > That might cause spikes in cpu/memory/disk usage that aren't ideal though. Yes, if there were a 100 remotes that's a bit of a fork-bomb. >> local sed_cmd= >> local remote >> # ref names and therefore remote names cannot contain '*?[]^' so we >> # only need to escape '.$/'. Using 'sort -r' means that if there is a >> # remote called "github" and another called "github/alice" we will try >> # and strip "github/alice" first. >> for remote in $(__git_remotes | sort -r) >> do >> remote="${remote//./\\./}" >> remote="${remote//\$/\\\$/}" >> remote="${remote//\//\\\//}" > > Just FYI since it took me hours to figure this out myself: I think this > would break tests on macos because of an old version of bash that > handles backslashes weirdly. I think removing the double quotes would > work around that issue, and be safe because word splitting doesn't > happen in assignments. Thanks, I'm not familiar with bash's extensions to parameter substitution. The completions can also but used under zsh (git-completion.zsh is pretty much abandoned I think) but it looks like bash and zsh agree on this expansion. >> sed_cmd="${sed_cmd} -e s/^refs\/remotes\/$remote\/// -e t" >> done > > Mostly just a note to myself if I end up using this idea: I think a > space in $remote would break this because bash would split up the arg to > sed. There's probably some way to fix that with extra escaping though? I should have said in the comment above that ref names cannot contain space, tab, newline etc. so we don't have to worry about the shell's word splitting splitting the patterns. They cannot contain backslashes either. >> __git for-each-ref --format="$fer_pfx$sfx" \ >> ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ >> "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | > > This would search for $cur_ in the wrong place because * only matches > one path component, right? Changing to ** might help, but then it would > match in places more it shouldn't. Oh, good point I'd missed that. Could we change sed to run with '-n' and generate an expression that looks like -e s|^refs/remotes/$remote/\($cur_\)|\1|p -e t with the appropriate escaping of $remote and $cur_ so that we only print the output of for-each-ref that matches $cur_? Best Wishes Phillip >> sed $sed_cmd | sort -u >> >> should work and means we're only forking three extra processes no >> matter how many remotes the user has. I'm not sure if it changes the >> output order when GIT_COMPLETION_IGNORE_CASE is set though. >> >> Best Wishes >> >> Phillip >
Op 2025-03-03 om 11:36 schreef phillip.wood123@gmail.com: > I'm not really sure what a "large" number looks like for people who > track a lot of repositories. On windows forking is pretty slow (our > shell-based test suite is incredibly slow on that platform) so I think > it makes sense to try and avoid adding new processes. Thanks, I didn't know about forking being slow on Windows. Do you know if we need to worry about hitting the max number or length of arguments to a command? https://pubs.opengroup.org/onlinepubs/9799919799/basedefs/limits.h.html under _POSIX_ARG_MAX says that at least 4KiB should be supported, which seems low enough to maybe cause trouble, but it's a minimum. https://www.in-ulm.de/~mascheck/various/argmax/ lists a bunch of values for different platforms, and from a brief scan it looks like all the remotely-modern platforms have values that are probably large enough for this. >> I'd like to do some testing to get actual performance numbers before >> trying to optimize this, because I think the optimization has some >> drawbacks, see below. >> >> If optimization is needed, another approach is to parallelize the forks: >> >> { >> local fer_pids= >> for ... >> do >> __git for-each-ref ... & >> fer_pids="$fer_pids $!" >> done >> test -z "$fer_pids" || wait $fer_pids >> } | sort | uniq -u >> >> That might cause spikes in cpu/memory/disk usage that aren't ideal >> though. > > Yes, if there were a 100 remotes that's a bit of a fork-bomb. I also realized after sending that email that it could run into issues with the number of available PIDs too. It could be improved by limiting to n child processes at a time and waiting for one to finish before starting the next, but that's more complication. I think your solution below is better. >>> local sed_cmd= >>> local remote >>> # ref names and therefore remote names cannot contain '*?[]^' so we >>> # only need to escape '.$/'. Using 'sort -r' means that if there is a >>> # remote called "github" and another called "github/alice" we will try >>> # and strip "github/alice" first. >>> for remote in $(__git_remotes | sort -r) >>> do >>> remote="${remote//./\\./}" >>> remote="${remote//\$/\\\$/}" >>> remote="${remote//\//\\\//}" >> >> Just FYI since it took me hours to figure this out myself: I think >> this would break tests on macos because of an old version of bash that >> handles backslashes weirdly. I think removing the double quotes would >> work around that issue, and be safe because word splitting doesn't >> happen in assignments. > > Thanks, I'm not familiar with bash's extensions to parameter > substitution. The completions can also but used under zsh (git- > completion.zsh is pretty much abandoned I think) but it looks like bash > and zsh agree on this expansion. The tests don't automatically run under zsh though, right? Maybe I should try to figure out how to do that in a separate patch... >>> sed_cmd="${sed_cmd} -e s/^refs\/remotes\/$remote\/// -e t" >>> done >> >> Mostly just a note to myself if I end up using this idea: I think a >> space in $remote would break this because bash would split up the arg >> to sed. There's probably some way to fix that with extra escaping though? > > I should have said in the comment above that ref names cannot contain > space, tab, newline etc. so we don't have to worry about the shell's > word splitting splitting the patterns. They cannot contain backslashes > either. Good to know. Is https://git-scm.com/docs/git-check-ref-format the place I should have read to understand what escaping is needed and what isn't because of the limitations on ref names? >>> __git for-each-ref --format="$fer_pfx$sfx" \ >>> ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ >>> "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | >> >> This would search for $cur_ in the wrong place because * only matches >> one path component, right? Changing to ** might help, but then it >> would match in places more it shouldn't. > > Oh, good point I'd missed that. Could we change sed to run with '-n' and > generate an expression that looks like > > -e s|^refs/remotes/$remote/\($cur_\)|\1|p -e t > > with the appropriate escaping of $remote and $cur_ so that we only print > the output of for-each-ref that matches $cur_? I think that would work. I was hoping to avoid sed, and I was thinking about using shell's ${parameter#word} to strip and test for matching prefixes. But I can't think of a way to do that with GIT_COMPLETION_IGNORE_CASE. Modern bash has ${parameter@L} to lowercase a string, but I don't think the version of bash on macos has that. With sed it would just be a matter of adding the i flag if needed. I don't think this is an issue, but do we need to worry about putting remote names and $cur_ on the command line where any other user on the same system can read them? $cur_ was already on the command line before this patch, and remote names appear on other git command lines all the time, so I doubt they're meant to be secret. Is a Helped-by trailer the right way to credit you for your suggestions on this?
On Sun, Mar 02, 2025 at 03:34:07PM -0500, David Mandelberg wrote: > Op 2025-03-02 om 09:17 schreef Phillip Wood: > > Hi David > > > > On 02/03/2025 07:45, David Mandelberg via GitGitGadget wrote: > > > From: David Mandelberg <david@mandelberg.org> > > > > > > Previously, some calls to for-each-ref passed fixed numbers of path > > > components to strip from refs, assuming that remote names had no slashes > > > in them. This made completions like: > > > > > > git push github/dseomn :com<Tab> > > > > > > Result in: > > > > > > git push github/dseomn :dseomn/completion-remote-slash > > > > > > With this patch, it instead results in: > > > > > > git push github/dseomn :completion-remote-slash > > > > This sounds like a useful improvement and I like the idea, but I think > > running "git for-each-ref" once for each remote is not going to scale > > very well for people who have a lot of remotes. I think it would be > > better to try and strip "refs/remote/$remote/" outside of "git for-each- > > ref". I've not tested it but I think something like > > Good point, I hadn't thought of that. Do you have a rough estimate of what > "a lot of remotes" is? 100ish, maybe? 5. In Git for Windows fork()-ing subshells and fork()+exec()-ing processes is rather costly, about an order of magnitude slower than on Linux. The rough equivalent of the body of your loop, with two subshells and a git process: time { a=$(echo 1) ; b=$(echo 2) ; git for-each-ref >/dev/null ; } takes on average 0.17s on a windows box I have access to (with fully packed refs, and merely 4 refs in total). So guess at about 4-5 remotes it would take over a second to react to my TABs... I would rather try to go in the opposite direction to see whether 'git for-each-ref' could be taught to strip the "refs/remote/$remote/" prefix with a format specifier option like '%(refname:strip=remote)'. That would surely be faster than any shell filtering we might come up with, and would also save us from the trouble of escaping glob and/or regex metacharacters for shell/sed pattern matching. > I'd like to do some testing to get > actual performance numbers before trying to optimize this, because I think > the optimization has some drawbacks, see below. > > If optimization is needed, another approach is to parallelize the forks: > > { > local fer_pids= > for ... > do > __git for-each-ref ... & > fer_pids="$fer_pids $!" > done > test -z "$fer_pids" || wait $fer_pids > } | sort | uniq -u > > That might cause spikes in cpu/memory/disk usage that aren't ideal though. Please don't do this, the completion script must not forkbomb the system.
David Mandelberg wrote: > I think that would work. I was hoping to avoid sed, and I was thinking about > using shell's ${parameter#word} to strip and test for matching prefixes. But > I can't think of a way to do that with GIT_COMPLETION_IGNORE_CASE. Modern > bash has ${parameter@L} to lowercase a string, but I don't think the version > of bash on macos has that. With sed it would just be a matter of adding the > i flag if needed. I haven't followed this discussion closely, but you can also use ${parameter,,} to do parameter case modification. Sadly, it looks like those were added in bash-4.0, so Apple will have kept their users from benefiting from that to avoid GPLv3 code. :/ I don't have an ancient bash-3.x like MacOS to test, but maybe they've backported some features? Though if you do end up using this, ${parameter,,} will work in a much wider range of bash version that ${parameter@L} which covers various releases in the OS's we aim to target (e.g. RHEL 8.x, which has bash-4.4).
Op 2025-03-03 om 14:33 schreef SZEDER Gábor: > On Sun, Mar 02, 2025 at 03:34:07PM -0500, David Mandelberg wrote: >> Good point, I hadn't thought of that. Do you have a rough estimate of what >> "a lot of remotes" is? 100ish, maybe? > > 5. > > In Git for Windows fork()-ing subshells and fork()+exec()-ing > processes is rather costly, about an order of magnitude slower than on > Linux. The rough equivalent of the body of your loop, with two > subshells and a git process: > > time { a=$(echo 1) ; b=$(echo 2) ; git for-each-ref >/dev/null ; } > > takes on average 0.17s on a windows box I have access to (with fully > packed refs, and merely 4 refs in total). So guess at about 4-5 > remotes it would take over a second to react to my TABs... Oh wow, that's much slower than I expected, thanks for the numbers! > I would rather try to go in the opposite direction to see whether 'git > for-each-ref' could be taught to strip the "refs/remote/$remote/" > prefix with a format specifier option like '%(refname:strip=remote)'. > > That would surely be faster than any shell filtering we might come up > with, and would also save us from the trouble of escaping glob and/or > regex metacharacters for shell/sed pattern matching. That doesn't fix the issue of needing to anchor the $cur_ part of the pattern after each refs/remote/$remote/ prefix. It might be possible to use something like %(refname:strip=remote) to find all remote branches with the remote stripped, then filter them for $cur_ in the shell or with sed though. What are the expectations around mixing versions of git itself and files in contrib? Can a single patch series add a feature to for-each-ref and use it in git-completion.bash, or does the feature in for-each-ref need to exist for some time before git-completion.bash can use it?
phillip.wood123@gmail.com writes: >> local fer_pids= >> for ... >> do >> __git for-each-ref ... & >> fer_pids="$fer_pids $!" >> done >> test -z "$fer_pids" || wait $fer_pids >> } | sort | uniq -u >> That might cause spikes in cpu/memory/disk usage that aren't ideal >> though. > > Yes, if there were a 100 remotes that's a bit of a fork-bomb. And they are all competing for a shared common resource for read access. I wonder if a single invocation of for-each-ref for refs/remotes/ hierarchy would be sufficient? That is, the calling shell script can set up a shell function "foo" and then eval $(for-each-ref --format='foo %(refname)' --shell refs/remotes/) and "foo" can do its thing avoiding as much forks as possible. As you are allowed bash-ism in completion script, you do not have to limit yourself to POSIX parameter substitution but can use substring expansion ${parameter:offset:length} and pattern substitution ${parameter/pattern/string} and other kind of bash exotics, as well as shell arrays (e.g., define an array __git__remote_branches[] for each remote, which may contain the remote-tracking branches for the remote).
On Mon, Mar 3, 2025 at 2:25 PM David Mandelberg <david@mandelberg.org> wrote: > > Op 2025-03-03 om 11:36 schreef phillip.wood123@gmail.com: > >>> local sed_cmd= > >>> local remote > >>> # ref names and therefore remote names cannot contain '*?[]^' so we > >>> # only need to escape '.$/'. Using 'sort -r' means that if there is a > >>> # remote called "github" and another called "github/alice" we will try > >>> # and strip "github/alice" first. > >>> for remote in $(__git_remotes | sort -r) > >>> do > >>> remote="${remote//./\\./}" > >>> remote="${remote//\$/\\\$/}" > >>> remote="${remote//\//\\\//}" > >> > >> Just FYI since it took me hours to figure this out myself: I think > >> this would break tests on macos because of an old version of bash that > >> handles backslashes weirdly. I think removing the double quotes would > >> work around that issue, and be safe because word splitting doesn't > >> happen in assignments. > > > > Thanks, I'm not familiar with bash's extensions to parameter > > substitution. The completions can also but used under zsh (git- > > completion.zsh is pretty much abandoned I think) but it looks like bash > > and zsh agree on this expansion. I don't think "abandoned" is the right characterization—at least, it's the completion script that Homebrew-installed Git users will get for Zsh, and it does re-use the Bash completion scripts (which therefore ought to stay portable between both if possible, otherwise we get stuff like 8776470cf3 (completion: repair config completion for Zsh, 2025-01-06)). The Zsh script has seen some fixups over time (such as my own 3c20acdf46 (completion: zsh: stop leaking local cache variable, 2024-04-30)). > > The tests don't automatically run under zsh though, right? Maybe I > should try to figure out how to do that in a separate patch... Correct—which is how bugs sneak in ;) I'm willing to manually test the patch if I can understand how to reproduce the issue—it sounds like having a remote name with a slash is sufficient? I started trying to test Zsh completion a while back when working on one of the patches mentioned above; I got as far as this hack [1], which is to say, not very far at all. [1]: https://github.com/git/git/commit/d8918195f18a503aa1a42fed3c66a0af8d04131f -- D. Ben Knoble
On Mon, Mar 3, 2025 at 2:43 PM Todd Zullinger <tmz@pobox.com> wrote: > > David Mandelberg wrote: > > I think that would work. I was hoping to avoid sed, and I was thinking about > > using shell's ${parameter#word} to strip and test for matching prefixes. But > > I can't think of a way to do that with GIT_COMPLETION_IGNORE_CASE. Modern > > bash has ${parameter@L} to lowercase a string, but I don't think the version > > of bash on macos has that. With sed it would just be a matter of adding the > > i flag if needed. > > I haven't followed this discussion closely, but you can > also use ${parameter,,} to do parameter case modification. > > Sadly, it looks like those were added in bash-4.0, so Apple > will have kept their users from benefiting from that to > avoid GPLv3 code. :/ > > I don't have an ancient bash-3.x like MacOS to test, but > maybe they've backported some features? I can confirm Apple's /bin/bash doesn't have ${x,,} > > Though if you do end up using this, ${parameter,,} will work > in a much wider range of bash version that ${parameter@L} > which covers various releases in the OS's we aim to target > (e.g. RHEL 8.x, which has bash-4.4). > > -- > Todd >
"D. Ben Knoble" <ben.knoble@gmail.com> writes: > On Mon, Mar 3, 2025 at 2:25 PM David Mandelberg <david@mandelberg.org> wrote: >> >> > Thanks, I'm not familiar with bash's extensions to parameter >> > substitution. The completions can also but used under zsh (git- >> > completion.zsh is pretty much abandoned I think) but it looks like bash >> > and zsh agree on this expansion. > > I don't think "abandoned" is the right characterization—at least, it's > the completion script that Homebrew-installed Git users will get for > Zsh, and it does re-use the Bash completion scripts (which therefore > ought to stay portable between both if possible, otherwise we get > stuff like 8776470cf3 (completion: repair config completion for Zsh, > 2025-01-06)). The Zsh script has seen some fixups over time (such as > my own 3c20acdf46 (completion: zsh: stop leaking local cache variable, > 2024-04-30)). Thanks for clarifying. >> The tests don't automatically run under zsh though, right? Maybe I >> should try to figure out how to do that in a separate patch... > > Correct—which is how bugs sneak in ;) > > I'm willing to manually test the patch if I can understand how to > reproduce the issue—it sounds like having a remote name with a slash > is sufficient? > > I started trying to test Zsh completion a while back when working on > one of the patches mentioned above; I got as far as this hack [1], > which is to say, not very far at all. Thanks.
"David Mandelberg via GitGitGadget" <gitgitgadget@gmail.com> writes: > This made completions like: > > git push github/dseomn :com<Tab> > > Result in: > > git push github/dseomn :dseomn/completion-remote-slash Apparently I am late to the party, but I have to wonder how prevalent it is to use such a name for remotes. Back in 2005 when I designed the remote namespaces and remote-tracking branches, the code may have been loose not to complain, but it certainly was not an intended way the remote names and remote-tracking hiearchies were to be used. Being able to say "refs/remote/*/master" and learn where everybody's 'master' branch points at with a single globbing was a powerful thing, but with refs/remotes/github/dseomn/master thrown into the mix, that would break down miserably. As a remote name, we still do use "is that a file on the filesystem?" to see if it is a local file:// URL (i.e. "git fetch github/dseomn" may be fetching from a subdirectory two levels down); in retrospect, we might have been better off if we said "well with a slash it must be a local directory and not remote nickname" to prevent such usage pattern. Anyway, this is not something we'd do lightly, but we may want to see if we should tighten the naming rules for remote nicknames and the mapping from nicknames to remote-tracking hiearchy, at Git 3.0 version boundary.
Hi Ben On 06/03/2025 20:24, D. Ben Knoble wrote: > On Mon, Mar 3, 2025 at 2:25 PM David Mandelberg <david@mandelberg.org> wrote: >> Op 2025-03-03 om 11:36 schreef phillip.wood123@gmail.com: >> >>> Thanks, I'm not familiar with bash's extensions to parameter >>> substitution. The completions can also but used under zsh (git- >>> completion.zsh is pretty much abandoned I think) but it looks like bash >>> and zsh agree on this expansion. > > I don't think "abandoned" is the right characterization—at least, it's > the completion script that Homebrew-installed Git users will get for > Zsh, and it does re-use the Bash completion scripts (which therefore > ought to stay portable between both if possible, otherwise we get > stuff like 8776470cf3 (completion: repair config completion for Zsh, > 2025-01-06)). The Zsh script has seen some fixups over time (such as > my own 3c20acdf46 (completion: zsh: stop leaking local cache variable, > 2024-04-30)). Thanks for clarifying that, I'd misunderstood the relationship between git-completion.zsh and git-completion.bash. Best Wishes Phillip
Op 2025-03-06 om 15:46 schreef Junio C Hamano: > Apparently I am late to the party, but I have to wonder how > prevalent it is to use such a name for remotes. From a quick search, it looks like I'm not the only one, but I don't know how common it is: https://github.com/desktop/desktop/issues/3618 I'm not sure if I'm remembering right, but I think I got the idea from seeing slashes in branch names to organize a hierarchy of branches, and realizing I could do the same with remotes. Was it intentional to allow slash in branch names, or is that a similar situation? > As a remote name, we still do use "is that a > file on the filesystem?" to see if it is a local file:// URL > (i.e. "git fetch github/dseomn" may be fetching from a subdirectory > two levels down); That's a good reason not to use slashes in remotes that I didn't know about, thanks! > Anyway, this is not something we'd do lightly, but we may want to > see if we should tighten the naming rules for remote nicknames and > the mapping from nicknames to remote-tracking hiearchy, at Git 3.0 > version boundary. What does that mean for this patch now? Do you want to fix the completion for the currently (unintentionally) supported remote names, or leave it broken to discourage their use? Fixing the completion might also give anybody reading the completion script the idea that slashes in remote names are somewhat sanctioned.
David Mandelberg <david@mandelberg.org> writes: > I'm not sure if I'm remembering right, but I think I got the idea from > seeing slashes in branch names to organize a hierarchy of branches, > and realizing I could do the same with remotes. Was it intentional to > allow slash in branch names, or is that a similar situation? Reference names were hierarchical pretty much from the beginning. >> Anyway, this is not something we'd do lightly, but we may want to >> see if we should tighten the naming rules for remote nicknames and >> the mapping from nicknames to remote-tracking hiearchy, at Git 3.0 >> version boundary. > > What does that mean for this patch now? Do you want to fix the > completion for the currently (unintentionally) supported remote names, > or leave it broken to discourage their use? Fixing the completion > might also give anybody reading the completion script the idea that > slashes in remote names are somewhat sanctioned. It probably would not mean much either way, and certainly not something we should worry about during pre-release feature freeze. The "fix" may take quite different shapes. Remote nick/name may still be allowed but the default fetch refspec left by "git clone" would start using the refs/remotes/nick-name/ hierarchy instead of the refs/remotes/nick/name hierarchy, for example, but as long as this round of completion fixes properly read the configuration to learn the remote-tracking hierarchy by reading the configured value of "remote.nick/name.fetch", such a change would not break the code in the completion script at all. Or remote nick/name may outright banned and people are encouraged to use nick-name instead, in which case even if the completion code is prepared to accept remote names with slashes in it, as long as it still works correctly with remote names with no slashes in it, it will keep working in such a future. Or the "fix" may be to declare "there is nothing to fix, if people want to use [remote "nick/name"] fetch = +refs/heads/*:refs/remotes/nickname/* [remote "nocknyme"] fetch = +refs/heads/*:refs/remotes/nock/nyme/* that's their choice, and the completion code would do the right thing anyway". I didn't read your patches, so the situation may be different and with any shape of "fix", or without any "fix", there might still be need for further polishing. But hopefully you got the idea. For this particular patch, there is nothing urgently needed. Well, we are right now in the pre-release feature freeze anyway ;-). Thanks.
Op 2025-03-07 om 15:57 schreef Junio C Hamano: > It probably would not mean much either way, and certainly not > something we should worry about during pre-release feature freeze. > > The "fix" may take quite different shapes. Remote nick/name may > still be allowed but the default fetch refspec left by "git clone" > would start using the refs/remotes/nick-name/ hierarchy instead of > the refs/remotes/nick/name hierarchy, for example, but as long as > this round of completion fixes properly read the configuration to > learn the remote-tracking hierarchy by reading the configured value > of "remote.nick/name.fetch", such a change would not break the code > in the completion script at all. Or remote nick/name may outright > banned and people are encouraged to use nick-name instead, in which > case even if the completion code is prepared to accept remote names > with slashes in it, as long as it still works correctly with remote > names with no slashes in it, it will keep working in such a future. > > Or the "fix" may be to declare "there is nothing to fix, if people > want to use > > [remote "nick/name"] fetch = +refs/heads/*:refs/remotes/nickname/* > [remote "nocknyme"] fetch = +refs/heads/*:refs/remotes/nock/nyme/* > > that's their choice, and the completion code would do the right thing > anyway". > > I didn't read your patches, so the situation may be different and > with any shape of "fix", or without any "fix", there might still be > need for further polishing. But hopefully you got the idea. For > this particular patch, there is nothing urgently needed. Well, we > are right now in the pre-release feature freeze anyway ;-). > > Thanks. Thanks for the explanation! > As a remote name, we still do use "is that a > file on the filesystem?" to see if it is a local file:// URL > (i.e. "git fetch github/dseomn" may be fetching from a subdirectory > two levels down) Btw, I just tested it, and I think this is an issue for remote names without slashes too: /tmp/tmp.sVAhhZuEzA$ git init foo Initialized empty Git repository in /tmp/tmp.sVAhhZuEzA/foo/.git/ /tmp/tmp.sVAhhZuEzA$ cd foo /tmp/tmp.sVAhhZuEzA/foo$ git init --bare origin Initialized empty Git repository in /tmp/tmp.sVAhhZuEzA/foo/origin/ /tmp/tmp.sVAhhZuEzA/foo$ git commit --allow-empty -m 'test' [main (root-commit) 24c92a9] test /tmp/tmp.sVAhhZuEzA/foo$ git push origin main Enumerating objects: 2, done. Counting objects: 100% (2/2), done. Writing objects: 100% (2/2), 162 bytes | 162.00 KiB/s, done. Total 2 (delta 0), reused 0 (delta 0), pack-reused 0 (from 0) To origin * [new branch] main -> main
David Mandelberg <david@mandelberg.org> writes: >> As a remote name, we still do use "is that a >> file on the filesystem?" to see if it is a local file:// URL >> (i.e. "git fetch github/dseomn" may be fetching from a subdirectory >> two levels down) > > Btw, I just tested it, and I think this is an issue for remote names > without slashes too: Yes, and that is an intended way local repositories are discovered and used. The problem with slashes is that the mistake surface becomes a lot larger. "ls<RETURN>" would show you "origin" immediately in the current working directory, but you wouldn't know if origin has origin/foo or origin/bar unless you look. Limiting the local filesystem reference to something like "only the ones that has no slashes, or begin with ../ or /" would make it less likely that you meant to push to somewhere and instead push to a local directory.
Op 2025-03-06 om 15:24 schreef D. Ben Knoble: > I'm willing to manually test the patch if I can understand how to > reproduce the issue—it sounds like having a remote name with a slash > is sufficient? Yup. I was able to reproduce it with these commands: /tmp/tmp.zOjfmdMx1i$ git init foo Initialized empty Git repository in /tmp/tmp.zOjfmdMx1i/foo/.git/ /tmp/tmp.zOjfmdMx1i$ cd foo /tmp/tmp.zOjfmdMx1i/foo$ git remote add with/slash /path/does/not/matter /tmp/tmp.zOjfmdMx1i/foo$ git commit -m 'test' --allow-empty [main (root-commit) 4b95a99] test /tmp/tmp.zOjfmdMx1i/foo$ git update-ref refs/remotes/with/slash/main 4b95a99 Then type: git push with/slash :ma<Tab>
diff --git a/contrib/completion/git-completion.bash b/contrib/completion/git-completion.bash index 17c044f7d60..5f4407be37f 100644 --- a/contrib/completion/git-completion.bash +++ b/contrib/completion/git-completion.bash @@ -807,11 +807,15 @@ __git_dwim_remote_heads () # employ the heuristic used by git checkout and git switch # Try to find a remote branch that cur_es the completion word # but only output if the branch name is unique - __git for-each-ref --format="$fer_pfx%(refname:strip=3)$sfx" \ - --sort="refname:strip=3" \ - ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ - "refs/remotes/*/$cur_*" "refs/remotes/*/$cur_*/**" | \ - uniq -u + local remote + for remote in $(__git_remotes); do + local base="$(__git_escape_fnmatch "refs/remotes/$remote")" + local strip="$(__git_count_path_components "$base")" + __git for-each-ref \ + --format="$fer_pfx%(refname:strip=$strip)$sfx" \ + ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ + "$base/$cur_*" "$base/$cur_*/**" + done | sort | uniq -u } # Lists refs from the local (by default) or from a remote repository. @@ -917,7 +921,8 @@ __git_refs () case "HEAD" in $match*|$umatch*) echo "${pfx}HEAD$sfx" ;; esac - __git for-each-ref --format="$fer_pfx%(refname:strip=3)$sfx" \ + local strip="$(__git_count_path_components "refs/remotes/$remote")" + __git for-each-ref --format="$fer_pfx%(refname:strip=$strip)$sfx" \ ${GIT_COMPLETION_IGNORE_CASE+--ignore-case} \ "refs/remotes/$remote/$match*" \ "refs/remotes/$remote/$match*/**" diff --git a/t/t9902-completion.sh b/t/t9902-completion.sh index 0219408358c..0c41ef9d5b1 100755 --- a/t/t9902-completion.sh +++ b/t/t9902-completion.sh @@ -149,7 +149,8 @@ fi test_expect_success 'setup for __git_find_repo_path/__gitdir tests' ' mkdir -p subdir/subsubdir && mkdir -p non-repo && - git init -b main otherrepo + git init -b main otherrepo && + git init -b main slashrepo ' test_expect_success '__git_find_repo_path - from command line (through $__git_dir)' ' @@ -681,6 +682,13 @@ test_expect_success 'setup for ref completion' ' ) && git remote add other "$ROOT/otherrepo/.git" && git fetch --no-tags other && + ( + cd slashrepo && + git commit --allow-empty -m initial && + git branch -m main branch/with/slash + ) && + git remote add remote/with/slash "$ROOT/slashrepo/.git" && + git fetch --no-tags remote/with/slash && rm -f .git/FETCH_HEAD && git init thirdrepo ' @@ -693,6 +701,8 @@ test_expect_success '__git_refs - simple' ' other/HEAD other/branch-in-other other/main-in-other + remote/with/slash/HEAD + remote/with/slash/branch/with/slash matching-tag EOF ( @@ -709,6 +719,8 @@ test_expect_success '__git_refs - full refs' ' refs/remotes/other/HEAD refs/remotes/other/branch-in-other refs/remotes/other/main-in-other + refs/remotes/remote/with/slash/HEAD + refs/remotes/remote/with/slash/branch/with/slash refs/tags/matching-tag EOF ( @@ -774,6 +786,19 @@ test_expect_success '__git_refs - configured remote' ' test_cmp expected "$actual" ' +test_expect_success '__git_refs - configured remote - with slash' ' + cat >expected <<-EOF && + HEAD + HEAD + branch/with/slash + EOF + ( + cur= && + __git_refs remote/with/slash >"$actual" + ) && + test_cmp expected "$actual" +' + test_expect_success '__git_refs - configured remote - full refs' ' cat >expected <<-EOF && HEAD @@ -916,17 +941,19 @@ test_expect_success '__git_refs - unique remote branches for git checkout DWIMer other/ambiguous other/branch-in-other other/main-in-other - remote/ambiguous - remote/branch-in-remote + remote/with/slash/HEAD + remote/with/slash/ambiguous + remote/with/slash/branch-in-remote + remote/with/slash/branch/with/slash matching-tag - HEAD branch-in-other branch-in-remote + branch/with/slash main-in-other EOF for remote_ref in refs/remotes/other/ambiguous \ - refs/remotes/remote/ambiguous \ - refs/remotes/remote/branch-in-remote + refs/remotes/remote/with/slash/ambiguous \ + refs/remotes/remote/with/slash/branch-in-remote do git update-ref $remote_ref main && test_when_finished "git update-ref -d $remote_ref" || return 1 @@ -946,6 +973,8 @@ test_expect_success '__git_refs - after --opt=' ' other/HEAD other/branch-in-other other/main-in-other + remote/with/slash/HEAD + remote/with/slash/branch/with/slash matching-tag EOF ( @@ -962,6 +991,8 @@ test_expect_success '__git_refs - after --opt= - full refs' ' refs/remotes/other/HEAD refs/remotes/other/branch-in-other refs/remotes/other/main-in-other + refs/remotes/remote/with/slash/HEAD + refs/remotes/remote/with/slash/branch/with/slash refs/tags/matching-tag EOF ( @@ -979,6 +1010,8 @@ test_expect_success '__git refs - excluding refs' ' ^other/HEAD ^other/branch-in-other ^other/main-in-other + ^remote/with/slash/HEAD + ^remote/with/slash/branch/with/slash ^matching-tag EOF ( @@ -995,6 +1028,8 @@ test_expect_success '__git refs - excluding full refs' ' ^refs/remotes/other/HEAD ^refs/remotes/other/branch-in-other ^refs/remotes/other/main-in-other + ^refs/remotes/remote/with/slash/HEAD + ^refs/remotes/remote/with/slash/branch/with/slash ^refs/tags/matching-tag EOF ( @@ -1022,6 +1057,8 @@ test_expect_success '__git_refs - do not filter refs unless told so' ' other/branch-in-other other/main-in-other other/matching/branch-in-other + remote/with/slash/HEAD + remote/with/slash/branch/with/slash matching-tag matching/tag EOF @@ -1142,6 +1179,8 @@ test_expect_success '__git_complete_refs - simple' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z matching-tag Z EOF ( @@ -1180,6 +1219,20 @@ test_expect_success '__git_complete_refs - remote' ' test_cmp expected out ' +test_expect_success '__git_complete_refs - remote - with slash' ' + sed -e "s/Z$//" >expected <<-EOF && + HEAD Z + HEAD Z + branch/with/slash Z + EOF + ( + cur= && + __git_complete_refs --remote=remote/with/slash && + print_comp + ) && + test_cmp expected out +' + test_expect_success '__git_complete_refs - track' ' sed -e "s/Z$//" >expected <<-EOF && HEAD Z @@ -1188,9 +1241,11 @@ test_expect_success '__git_complete_refs - track' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z matching-tag Z - HEAD Z branch-in-other Z + branch/with/slash Z main-in-other Z EOF ( @@ -1235,6 +1290,8 @@ test_expect_success '__git_complete_refs - suffix' ' other/HEAD. other/branch-in-other. other/main-in-other. + remote/with/slash/HEAD. + remote/with/slash/branch/with/slash. matching-tag. EOF ( @@ -1260,6 +1317,20 @@ test_expect_success '__git_complete_fetch_refspecs - simple' ' test_cmp expected out ' +test_expect_success '__git_complete_fetch_refspecs - with slash' ' + sed -e "s/Z$//" >expected <<-EOF && + HEAD:HEAD Z + HEAD:HEAD Z + branch/with/slash:branch/with/slash Z + EOF + ( + cur= && + __git_complete_fetch_refspecs remote/with/slash && + print_comp + ) && + test_cmp expected out +' + test_expect_success '__git_complete_fetch_refspecs - matching' ' sed -e "s/Z$//" >expected <<-EOF && branch-in-other:branch-in-other Z @@ -1340,8 +1411,8 @@ test_expect_success '__git_complete_worktree_paths with -C' ' test_expect_success 'git switch - with no options, complete local branches and unique remote branch names for DWIM logic' ' test_completion "git switch " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1487,8 +1558,8 @@ test_expect_success 'git-bisect - existing view subcommand is recognized and ena test_expect_success 'git checkout - completes refs and unique remote branches for DWIM' ' test_completion "git checkout " <<-\EOF HEAD Z - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1496,6 +1567,8 @@ test_expect_success 'git checkout - completes refs and unique remote branches fo other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1515,8 +1588,8 @@ test_expect_success 'git switch - with GIT_COMPLETION_CHECKOUT_NO_GUESS=1, compl test_expect_success 'git switch - --guess overrides GIT_COMPLETION_CHECKOUT_NO_GUESS=1, complete local branches and unique remote names for DWIM logic' ' GIT_COMPLETION_CHECKOUT_NO_GUESS=1 test_completion "git switch --guess " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1525,8 +1598,8 @@ test_expect_success 'git switch - --guess overrides GIT_COMPLETION_CHECKOUT_NO_G test_expect_success 'git switch - a later --guess overrides previous --no-guess, complete local and remote unique branches for DWIM' ' test_completion "git switch --no-guess --guess " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1549,14 +1622,16 @@ test_expect_success 'git checkout - with GIT_COMPLETION_NO_GUESS=1 only complete other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' test_expect_success 'git checkout - --guess overrides GIT_COMPLETION_NO_GUESS=1, complete refs and unique remote branches for DWIM' ' GIT_COMPLETION_CHECKOUT_NO_GUESS=1 test_completion "git checkout --guess " <<-\EOF HEAD Z - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1564,6 +1639,8 @@ test_expect_success 'git checkout - --guess overrides GIT_COMPLETION_NO_GUESS=1, other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1576,14 +1653,16 @@ test_expect_success 'git checkout - with --no-guess, only completes refs' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' test_expect_success 'git checkout - a later --guess overrides previous --no-guess, complete refs and unique remote branches for DWIM' ' test_completion "git checkout --no-guess --guess " <<-\EOF HEAD Z - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1591,6 +1670,8 @@ test_expect_success 'git checkout - a later --guess overrides previous --no-gues other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1603,6 +1684,8 @@ test_expect_success 'git checkout - a later --no-guess overrides previous --gues other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1616,6 +1699,8 @@ test_expect_success 'git checkout - with checkout.guess = false, only completes other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1623,8 +1708,8 @@ test_expect_success 'git checkout - with checkout.guess = true, completes refs a test_config checkout.guess true && test_completion "git checkout " <<-\EOF HEAD Z - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1632,6 +1717,8 @@ test_expect_success 'git checkout - with checkout.guess = true, completes refs a other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1639,8 +1726,8 @@ test_expect_success 'git checkout - a later --guess overrides previous checkout. test_config checkout.guess false && test_completion "git checkout --guess " <<-\EOF HEAD Z - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -1648,6 +1735,8 @@ test_expect_success 'git checkout - a later --guess overrides previous checkout. other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1661,6 +1750,8 @@ test_expect_success 'git checkout - a later --no-guess overrides previous checko other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1673,6 +1764,8 @@ test_expect_success 'git switch - with --detach, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1685,6 +1778,8 @@ test_expect_success 'git checkout - with --detach, complete only references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1857,6 +1952,8 @@ test_expect_success 'git switch - with -d, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1869,6 +1966,8 @@ test_expect_success 'git checkout - with -d, complete only references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1877,11 +1976,15 @@ test_expect_success 'git switch - with --track, complete only remote branches' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF test_completion "git switch -t " <<-\EOF other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1890,11 +1993,15 @@ test_expect_success 'git checkout - with --track, complete only remote branches' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF test_completion "git checkout -t " <<-\EOF other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1914,6 +2021,8 @@ test_expect_success 'git checkout - with --no-track, complete only local referen other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1926,6 +2035,8 @@ test_expect_success 'git switch - with -c, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1938,6 +2049,8 @@ test_expect_success 'git switch - with -C, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1950,6 +2063,8 @@ test_expect_success 'git switch - with -c and --track, complete all references' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1962,6 +2077,8 @@ test_expect_success 'git switch - with -C and --track, complete all references' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1974,6 +2091,8 @@ test_expect_success 'git switch - with -c and --no-track, complete all reference other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1986,6 +2105,8 @@ test_expect_success 'git switch - with -C and --no-track, complete all reference other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -1998,6 +2119,8 @@ test_expect_success 'git checkout - with -b, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2010,6 +2133,8 @@ test_expect_success 'git checkout - with -B, complete all references' ' other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2022,6 +2147,8 @@ test_expect_success 'git checkout - with -b and --track, complete all references other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2034,6 +2161,8 @@ test_expect_success 'git checkout - with -B and --track, complete all references other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2046,6 +2175,8 @@ test_expect_success 'git checkout - with -b and --no-track, complete all referen other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2058,13 +2189,15 @@ test_expect_success 'git checkout - with -B and --no-track, complete all referen other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' test_expect_success 'git switch - for -c, complete local branches and unique remote branches' ' test_completion "git switch -c " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2073,8 +2206,8 @@ test_expect_success 'git switch - for -c, complete local branches and unique rem test_expect_success 'git switch - for -C, complete local branches and unique remote branches' ' test_completion "git switch -C " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2111,8 +2244,8 @@ test_expect_success 'git switch - for -C with --no-track, complete local branche test_expect_success 'git checkout - for -b, complete local branches and unique remote branches' ' test_completion "git checkout -b " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2121,8 +2254,8 @@ test_expect_success 'git checkout - for -b, complete local branches and unique r test_expect_success 'git checkout - for -B, complete local branches and unique remote branches' ' test_completion "git checkout -B " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2159,8 +2292,8 @@ test_expect_success 'git checkout - for -B with --no-track, complete local branc test_expect_success 'git switch - with --orphan completes local branch names and unique remote branch names' ' test_completion "git switch --orphan " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2175,8 +2308,8 @@ test_expect_success 'git switch - --orphan with branch already provided complete test_expect_success 'git checkout - with --orphan completes local branch names and unique remote branch names' ' test_completion "git checkout --orphan " <<-\EOF - HEAD Z branch-in-other Z + branch/with/slash Z main Z main-in-other Z matching-branch Z @@ -2192,6 +2325,8 @@ test_expect_success 'git checkout - --orphan with branch already provided comple other/HEAD Z other/branch-in-other Z other/main-in-other Z + remote/with/slash/HEAD Z + remote/with/slash/branch/with/slash Z EOF ' @@ -2206,7 +2341,8 @@ test_expect_success 'git restore completes modified files' ' test_expect_success 'teardown after ref completion' ' git branch -d matching-branch && git tag -d matching-tag && - git remote remove other + git remote remove other && + git remote remove remote/with/slash '