Hi, On Sat, 14 Jun 2008, Allan Wind wrote: > Is anyone looking into fixing gc so it does not change permission on > files? I would think it should consider current permissions of files > before umask. > > I believe Nigel reported this issue with: > http://thread.gmane.org/gmane.comp.version-control.git/79213/focus=79237 > and still seeing issues with 1.5.5.4. It would have been nice to briefly illustrate the problem again. In the message you referred to, it seems to suggest that git-gc (in fact, git pack-refs) set the permissions wrong in a shared repository. However, I cannot reproduce at all: -- snip -- [PATCH] Verify that permissions for packed-refs are set correctly Signed-off-by: Johannes Schindelin <johannes.schindelin@xxxxxx> --- t/t1301-shared-repo.sh | 18 ++++++++++++++++++ 1 files changed, 18 insertions(+), 0 deletions(-) diff --git a/t/t1301-shared-repo.sh b/t/t1301-shared-repo.sh index 5e4252a..c4d3239 100755 --- a/t/t1301-shared-repo.sh +++ b/t/t1301-shared-repo.sh @@ -43,6 +43,24 @@ test_expect_success 'update-server-info honors core.sharedRepository' ' esac ' +test_expect_success 'pack-refs honors core.sharedRepository' ' + + (umask 0217 && + : > a2 && + git pack-refs) && + actual="$(ls -l .git/packed-refs)" && + case "$actual" in + -r--r--r--*) + : happy + ;; + *) + echo Oops, .git/packed-refs is not 0444 + false + ;; + esac + +' + for u in 0660:rw-rw---- \ 0640:rw-r----- \ 0600:rw------- \ -- snap -- This succeeds. Indeed, it should succeed, since the relevant part in lockfile.c was fixed in version v1.4.0~4. However, I suspect from the original message that the reflogs did not have the correct permissions. But again, a simple test succeeds: -- snip -- [PATCH] Add a test for logrefs in bare repositories Signed-off-by: Johannes Schindelin <johannes.schindelin@xxxxxx> --- t/t1301-shared-repo.sh | 36 ++++++++++++++++++++++++++++++++++++ 1 files changed, 36 insertions(+), 0 deletions(-) diff --git a/t/t1301-shared-repo.sh b/t/t1301-shared-repo.sh index 5e4252a..8530a56 100755 --- a/t/t1301-shared-repo.sh +++ b/t/t1301-shared-repo.sh @@ -24,6 +24,39 @@ test_expect_success 'shared=all' ' test 2 = $(git config core.sharedrepository) ' +test_expect_success 'reflogs honor core.sharedRepository' ' + git config core.logAllRefUpdates true && + echo 2 > a1 && + git add a1 && + test_tick && + git commit -m a1 && + actual="$(ls -l .git/logs/refs/heads/master)" && + case "$actual" in + -rw-rw-r--*) + : happy + ;; + *) + echo Oops, .git/logs/refs/heads/master is not 0664 + false + ;; + esac && + (mkdir bare.git && + cd bare.git && + git --bare init --shared && + git config core.logAllRefUpdates true) && + git push bare.git master:refs/heads/new && + actual="$(ls -l bare.git/logs/refs/heads/new)" && + case "$actual" in + -rw-rw-*) + : happy + ;; + *) + echo Oops, bare.git/logs/refs/heads/new is not 0664 + false + ;; + esac +' + test_expect_success 'update-server-info honors core.sharedRepository' ' : > a1 && git add a1 && -- 1.5.5.1.780.g346de -- snap -- So now I got curious and tested it with v1.5.4.4, and indeed, it failed. If you are curious who/what fixed it, go ahead, bisect it. I wrote all this down so that the next time, you could verify yourself if the bug was fixed or not, instead of having to ask here. Ciao, Dscho -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
