Re: [PATCH v4 00/12] Wildmatch v4

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 12, 2012, 17:05 UTC
Message-ID: <7vbog74o59.fsf@alter.siamese.dyndns.org>
In-Reply-To: <507848DF.6060301@web.de>

Torsten Bögershausen <tboegi@web.de> writes:

Show 49 quoted lines

> On 11.10.12 13:56, Nguyen Thai Ngoc Duy wrote:
>> On Thu, Oct 11, 2012 at 11:33 AM, Junio C Hamano <gitster@pobox.com> wrote:
>>> Junio C Hamano <gitster@pobox.com> writes:
>>>
>>>> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>>>>
>>>>> Really small updates. I did not want to resend it this soon but this
>>>>> may fix the compile errors for Junio.
>>>
>>> t3070 seems to break TAP,
>>>
>>>     *** prove ***
>>>     t3070-wildmatch.sh .. Failed 1/151 subtests
>>>
>>>     Test Summary Report
>>>     -------------------
>>>     t3070-wildmatch.sh (Wstat: 0 Tests: 150 Failed: 0)
>>>       Parse errors: Tests out of sequence.  Found (76) but expected (75)
>>>                     Tests out of sequence.  Found (77) but expected (76)
>>>                     Tests out of sequence.  Found (78) but expected (77)
>>>                     Tests out of sequence.  Found (79) but expected (78)
>>>                     Tests out of sequence.  Found (80) but expected (79)
>>>     Displayed the first 5 of 77 TAP syntax errors.
>>>
>>> This probably is due to this part of the output:
>>>
>>>     ok 72 - wildmatch 1 1 [ab] [\[:]ab]
>>>     ok 73 - wildmatch 1 1 ?a?b \??\?b
>>>     ok 74 - wildmatch 1 1 abc ^G^Hok 75 - wildmatch 0 0 foo
>>>     ok 76 - wildmatch 1 0 foo/bar/baz/to **/t[o]
>>>     ok 77 - wildmatch 1 1 a1B [[:alpha:]][[:digit:]][[:upper:]]
>>>
>> 
>> It seems to prove fine here with perl 5.12.3, Test-Harness-3.230,
>> bash-4.0_p38. What version do you use?
>> 
> Some problems even here (Mac OS) commit 2aeb6d4d7884f4c4425
>
> not ok 61 - wildmatch 0 0 \ \
> not ok 62 - wildmatch 0 0 /\ */\
> not ok 69 - wildmatch 1 1 [ab] [[:]ab]
> not ok 71 - wildmatch 1 1 [ab] [[:digit]ab]
> not ok 80 - wildmatch 1 0 1 [[:digit:][:upper:][:space:]]
> not ok 81 - wildmatch 0 0 1 [[:digit:][:upper:][:spaci:]]
> not ok 88 - wildmatch 1 0 _ [[:alnum:][:alpha:][:blank:][:cntrl:][:digit:][:graph:][:lower:][:print:][:punct:][:space:][:upper:][:xdigit:]]
> not ok 89 - wildmatch 1 0 _ [[:alnum:][:alpha:][:blank:][:cntrl:][:digit:][:graph:][:lower:][:print:][:punct:][:space:][:upper:][:xdigit:]]
> not ok 93 - wildmatch 1 1 y [a-c[:digit:]x-z]
>
> I have 2 perls,...

The problem under discussion was that the output from passing tests were mangled and not interpreted correctly by a program that analyzes them, which happens to be written in Perl. Even in that context, it was pointless to ask for Perl versions, as it was clear that the input to that Perl-written program, i.e. output from the tests, were not formatted correctly to say "ok NN message".

If you are getting "not ok", I do not think you need to look at your Perl. The wildmatch tests are failing, and it is far more likely that the wildmatch code is failing in your environment, than the test harness that runs wildmatch code is failing.

Neither the wildmatch code itself (in C), nor the test harness that is giving you the above "not ok" (in shell) depends on Perl.

Thanks for a report, but the debugging needs to see how the matching code in mildmatch misbehaves.

I suspect it is another case where the platform fnmatch (from MacOS) is not behaving as these tests expect.

Nguyen, how about updating the match () shell function in 3070 so that it not just says not-ok, but indicates what failed (wildmatch failed, or wildmatch passed but fnmatch failed), at least when the test is run as "./t3070-*.sh -v -i"?