From: Christian Couder <chriscool@tuxfamily.org>
Date: Mon, 11 Feb 2008 04:55:19 GMT
Subject: Re: [PATCH] diff.c: guard config parser from value=NULL
Message-ID: <200802110555.19383.chriscool@tuxfamily.org>
In-Reply-To: <20080210190145.GB5129@artemis.madism.org>

Le dimanche 10 février 2008, Pierre Habouzit a écrit :
> On Sun, Feb 10, 2008 at 06:35:34PM +0000, Christian Couder wrote:
> > +static int xstrdup_confval(const char *dest, const char *var, const
> > char *value)
>
>   Why not calling that "git_config_string" in config.[hc] and be
> API-consistent ?

In this function, there is:

if (!value || !*value)
	return error("%s: lacks value", var);

but in some places an empty string is not an error, and in some other places 
we "die" in case of error instead of using "return error..."
So I am not sure this function can be generally usefull as it is now.

Maybe I should add a flag parameter to control if we should "die" or "return 
error" and if we should accept an empty string or not.

What do you think ?

Christian.