From: Michael J Gruber <hidden> Date: 2016-06-15 22:46:06
"git notes show" barfs when there is no note to show. This introduces a
test (yes, a test!) and, in a second round, reacts more gracefully to
empty notes and adjust the expected test output accordingly.
Note that in both cases (before/after the patch) the return code is
non-zero: It's 128 in the ungraceful case, 1 when "dieing gracefully",
uhm...
Michael J Gruber (2):
git notes show: test empty notes
handle empty notes gracefully
git-notes.sh | 2 ++
t/t3301-notes.sh | 5 +++++
2 files changed, 7 insertions(+), 0 deletions(-)
From: Michael J Gruber <hidden> Date: 2016-06-15 22:46:06
Currently, git-notes barfs when asked to show an empty (i.e.
non-existing) note. Change this to explicitely say there is none.
---
git-notes.sh | 2 ++
t/t3301-notes.sh | 2 +-
2 files changed, 3 insertions(+), 1 deletions(-)
From: Michael J Gruber <hidden> Date: 2016-06-15 22:46:07
Michael J Gruber venit, vidit, dixit 06.02.2009 16:19:
"git notes show" barfs when there is no note to show. This introduces a
test (yes, a test!) and, in a second round, reacts more gracefully to
empty notes and adjust the expected test output accordingly.
Note that in both cases (before/after the patch) the return code is
non-zero: It's 128 in the ungraceful case, 1 when "dieing gracefully",
uhm...
Michael J Gruber (2):
git notes show: test empty notes
handle empty notes gracefully
git-notes.sh | 2 ++
t/t3301-notes.sh | 5 +++++
2 files changed, 7 insertions(+), 0 deletions(-)
Uhm, I'm about 1 hour late with my afternoon coffee, and I'm afraid that
shows:
1) I meant "bark", not "barf", although both make sense here.
2) When did "format.signoff = true" in config stop working? OK, it never
did. So please consider this (anyway trivial series):
Signed-off-by: Michael J Gruber <redacted>
From: Johannes Schindelin <hidden> Date: 2016-06-15 22:46:07
Hi,
On Fri, 6 Feb 2009, Michael J Gruber wrote:
quoted hunk
Currently, git-notes barfs when asked to show an empty (i.e.
non-existing) note. Change this to explicitely say there is none.
---
git-notes.sh | 2 ++
t/t3301-notes.sh | 2 +-
2 files changed, 3 insertions(+), 1 deletions(-)
Completely forgot to mention that I think you want to use test_must_fail
here. And maybe you want to be more explicit, by specifying which
commit's notes are expected not to be there.
We would not want the test to succeed for all the wrong reasons, would we?
Ciao,
Dscho
Yes. If "git notes show" returns 0 then even better. It does neither
before nor after the patch. Or should we always expect 1 and test for
that? I thought about grepping the output for "fatal" (which appears
now) but that seemed ugly.
Michael
From: Johannes Sixt <hidden> Date: 2016-06-15 22:46:07
Johannes Schindelin schrieb:
On Fri, 6 Feb 2009, Michael J Gruber wrote:
quoted
+test_expect_success 'handle empty notes gracefully' '
git notes show || test 1 = $?
Completely forgot to mention that I think you want to use test_must_fail
here.
Wouldn't work. We want to differentiate between different exit codes. But
after test_must_fail the exit code is gone. In this simple case this
should work:
git notes show; test 1 = $?
-- Hannes
From: Michael J Gruber <hidden> Date: 2016-06-15 22:46:07
Johannes Schindelin venit, vidit, dixit 06.02.2009 16:38:
Hi,
On Fri, 6 Feb 2009, Michael J Gruber wrote:
quoted
Currently, git-notes barfs when asked to show an empty (i.e.
non-existing) note. Change this to explicitely say there is none.
---
git-notes.sh | 2 ++
t/t3301-notes.sh | 2 +-
2 files changed, 3 insertions(+), 1 deletions(-)
Completely forgot to mention that I think you want to use test_must_fail
here.
What does test_must_fail mean? I don't see it in t/README.
And maybe you want to be more explicit, by specifying which
commit's notes are expected not to be there.
Well, HEAD's notes. I can say HEAD explicitly, of course.
We would not want the test to succeed for all the wrong reasons, would we?
Well, I could test for the friendly "No note for commit" message on
output, I just thought that is fragile.
You see, I did not even want to write a test for such a simple patch... ;)
OK, do we agree on the following intended behaviour for git notes show:
- return 0 if a note can be shown
- return 1 if there is none (i.e. die gracefully)
- return something else (i.e. die fatally) if something really bad happens
Then I should rewrite the test to check for "1" and only "1".
Now, the coffee...
Michael