tests failing on Cygwin

Hallo,

I couldn't find a better place to report this and i don't know whether anyone really cares about this, but i'll try anyway.

I want to make a few corrections to the core documentation. I cloned git.git and before starting the actual i built it and ran `make test', just in case, and some tests failed.

I am doing this on XP/Cygwin. Some of them may be irrelevant for Cygwin; if so, they probably should be skipped. And maybe i am just misunderstanding something.

Here's the list of files and names of failed tests:

t0001-init.sh:
* init notices EPERM

t1004-read-tree-m-u-wf.sh:
* funny symlink in work tree, un-unlink-able

t3300-funny-names.sh:
* git diff-tree delete with-funny
* ... and all the test from here until the end of that file

t3301-notes.sh:
* test_expect_success 'git notes copy --for-rewrite (disabled)' '
        git config notes.rewrite.bar false &&
        echo $(git rev-parse HEAD~3) $(git rev-parse HEAD) |
        git notes copy --for-rewrite=bar &&
        git log -2 > output &&
        test_cmp expect output
    '

t556x_common:
* direct refs/heads/master not found
* no export by default
* static file if http.getanyfile true is ok
* http.uploadpack false
* http.receivepack default disabled
* http.receivepack false

t7300-clean.sh:
* removal failure

t7606-merge-custom.sh:
* merge c2 with a custom strategy

On Sat, Aug 7, 2010 at 21:37, Amir E. Aharoni <amir.aharoni@mail.huji.ac.il> wrote:

Did you run these tests as root? The 'init notices EPERM' test at least seems to have failed due to that.

I have a pending patch series (on list) to make the test suite not fail when you run it as the root user.

It looks like you running them as the superuser might not explain all of them though, but I didn't go through them all in detail, and I don't have a cygwin system.

It also helps you run the failing tests like this:

    ./failing-test.sh --verbose --debug

And provide the full output from that. The output from "uname -a" would also be helpful.

On Sat, Aug 7, 2010 at 21:37, Amir E. Aharoni <amir.aharoni@mail.huji.ac.il> wrote:

(forgot to reply to this)

This is the right place, and we care. But we don't have a lot of people using Windows here, so stuff like this might be broken.

The Windows builds of Git use MINGW now that I can tell, not Cygwin, so Cygwin might break in some ways that MINGW doesn't.

Anyway, un-breaking it would be nice.