From: Junio C Hamano <gitster@pobox.com>
Date: Mon, 03 Sep 2012 01:53:13 GMT
Subject: Re: Test failures in t4034
Message-ID: <7vehmjg9di.fsf@alter.siamese.dyndns.org>
In-Reply-To: <5042494D.9040401@ramsay1.demon.co.uk>

Ramsay Jones <ramsay@ramsay1.demon.co.uk> writes:

> Yes, there was a net increase in the line count when I introduced
> die(), but the main program flow was less cluttered by error handling.
> The net result looked much better, so I thought it was worth it.
>
> What may not be too obvious, however, is that test-regex.c was written
> to be independent of git.

That part I was very aware of actually; it it is a bit tricky to
tell what the right thing to do, though.  Your test itself needs to
be pretty much portable without the portability help you would get
git-compat-util.h, but the point of this kind of test is to tell if
you want to define preprocessor macros that may affect the behaviour
of such compatibility layer ;-)

> Given that I'm now building it as part of git, I should have simply
> #included <git-compat-util.h> and used the die() routine from libgit.a
> (since I'm now *relying* on test-regex being linked with libgit.a).

OK.

>>> +int main(int argc, char **argv)
>>> +{
>>> +	char *pat = "[^={} \t]+";
>>> +	char *str = "={}\nfred";
>>> +	regex_t r;
>>> +	regmatch_t m[1];
>>> +
>>> +	if (regcomp(&r, pat, REG_EXTENDED | REG_NEWLINE))
>>> +		die("failed regcomp() for pattern '%s'", pat);
>>> +	if (regexec(&r, str, 1, m, 0))
>>> +		die("no match of pattern '%s' to string '%s'", pat, str);
>>> +
>>> +	/* http://sourceware.org/bugzilla/show_bug.cgi?id=3957  */
>>> +	if (m[0].rm_so == 3) /* matches '\n' when it should not */
>>> +		exit(1);
>> 
>> This could be the third call site of die() that tells the user to
>> build with NO_REGEX=1.  Then "cd t && sh t0070-fundamental.sh -i -v" would
>> give that message directly to the user.
>
> Hmm, even without "-i -v", it's *very* clear what is going on, but sure
> it wouldn't hurt either. (Also, I wanted to be able to distinguish an exit
> via die() from a "test failed" error return).

OK.

Thanks.