On Sun, Sep 14, 2014 at 06:20:57PM +0200, Fabian Ruch wrote:
> Hi David,
>
> On 09/14/2014 10:30 AM, David Aguilar wrote:
> > Ensure that rev-parse --verify --quiet is silent when asked
> > about deleted reflog entries.
> >
> > Signed-off-by: David Aguilar <davvid@xxxxxxxxx>
> > ---
> > This verifies and depends on "refs: make rev-parse --quiet actually quiet".
> >
> > t/t1503-rev-parse-verify.sh | 9 +++++++++
> > 1 file changed, 9 insertions(+)
> >
> > diff --git a/t/t1503-rev-parse-verify.sh b/t/t1503-rev-parse-verify.sh
> > index 813cc1b..731c21c 100755
> > --- a/t/t1503-rev-parse-verify.sh
> > +++ b/t/t1503-rev-parse-verify.sh
> > @@ -83,6 +83,15 @@ test_expect_success 'fails silently when using -q' '
> > test -z "$(cat error)"
> > '
> >
> > +test_expect_success 'fails silently when using -q with deleted reflogs' '
> > + ref=$(git rev-parse HEAD) &&
> > + : >.git/logs/refs/test &&
> > + git update-ref -m test refs/test "$ref" &&
>
> I'm just curious, why not simply
>
> git branch test
> ?
Maybe it's a bad reason, but I wanted to replicate the behavior
that git stash expects -- it writes to a ref outside of
refs/heads/. I thought it'd be good to exercise that same
machinery since it will involve different code paths.
> > + git reflog delete --updateref --rewrite refs/test@{0} &&
> > + test_must_fail git rev-parse --verify --quiet refs/test@{0} 2>error &&
>
> Is it a shortcoming of the specification that it doesn't consider
> whatever might be written to stdout? Is it acceptable that if the
> git-rev-parse command succeeds, the error message from test_must_fail
> will be written to the file "error" and, therefore, somewhat hidden from
> the user running the tests?
Good point. The --quiet spec doesn't say anything about stdout,
but for this test it probably wouldn't hurt to capture both
stdout and stderr and assert emptiness.
I can reroll this patch so that 2>error becomes >error 2>&1.
> > + test -z "$(cat error)"
>
> test(1) comes with an option (-s) to perform such tests and test-lib.sh
> defines test_must_be_empty which additionally outputs the given file's
> contents if its not empty.
test_must_be_empty would be a good fit here. That said, none of
the other tests in this file use test_must_be_empty.
It might be worth doing a follow-up patch that converts all of the
tests in this file to use test_must_be_empty instead of
test -z "$(cat error)". I'll reroll.
Thanks,
--
David
--
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