Re: [PATCH] [Outreachy] git/userdiff.c fix regex pattern error

From: Junio C Hamano <gitster@pobox.com>
Date: Oct 12, 2018, 08:41 UTC
Message-ID: <xmqqftxbblpl.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <nycvar.QRO.7.76.6.1810120955090.45@tvgsbejvaqbjf.bet>

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

Show 10 quoted lines

> On Sun, 7 Oct 2018, Junio C Hamano wrote:
>
>> And then there is an unnamed misdesigned language that has a
>> regmatch() function, which takes a string that contains a regular
>> expression, but somehow requires that string to begin and end with a
>> slash for no justifiable reason ;-).
>
> It makes more sense once you realize that /<regexp>/ is a very nice
> syntactic construct to obtain an instance of a regular expression object,
> rather than a string describing one.

In Perl, qr/<regexp>/ is often how you use that syntactic construct.

The unnamed misdesigned language literally uses a string whose content is /<regexp>/; iow,

	preg_match('/a.*b/', ...)

is what you see, and there is no "regex object involved. As far as the function is concerned, it is receiving a string, that represents a regexp. There is no reason to require the pattern _string_ to begin and end with slash, but somehow they do.

> But in C, we do not have objects, so the way to describe a regular
> expression is a string (which you have to compile into an opaque data
> structure before applying it).

Absolutely. C does not have regexp object literal and the literal you write to represent a regexp is just a string that you would pass to functions like regcomp(). But at least C is not so brain-dead to require slashes on either end of that string ;-)