Re: [PATCH] Use setenv(), fix warnings

[Timo Hirvonen <tihirvon@gmail.com>]

On Sun, 26 Feb 2006 10:06:41 -0800
Junio C Hamano <junkio@cox.net> wrote:

> Timo Hirvonen <tihirvon@gmail.com> writes:
> 
> >   - Use setenv() instead of putenv()
> >   - Fix -Wundef -Wold-style-definition warnings
> >   - Make pll_free() static
> 
> I think the last one makes sense, and I can see why some people
> may prefer -Wundef but I am not sure about the first one.  Care
> to defend why we should prefer setenv()?  IIRC, initially we did
> not use setenv() anywhere because certain platforms only had
> putenv().

I was confused by putenv(3) man page.  I thought it wanted malloc'ed
strings (no 'const' in the parameter -> warning).  It appears that
statically allocated strings are accepted but _automatic_ variables
aren't.  I noticed setenv is now in compat/ so I though it was good idea
to use it.

Sorry for the noise.

> > diff --git a/fsck-objects.c b/fsck-objects.c
> > @@ -483,7 +483,7 @@ int main(int argc, char **argv)
> >  	if (standalone && check_full)
> >  		die("Only one of --standalone or --full can be used.");
> >  	if (standalone)
> > -		putenv("GIT_ALTERNATE_OBJECT_DIRECTORIES=");
> > +		setenv("GIT_ALTERNATE_OBJECT_DIRECTORIES", "", 1);
> 
> For platforms with only putenv we did this; here, what we really
> wanted to do was unsetenv.

putenv(3):
    "If the argument `string` is of the form `name`, and does not
    contain an `=' character, then the variable `name` is removed from
    the environment."

So the variable is emptied, not removed.  But usually empty environment
variables are treated as if they didn't exist...

-- 
http://onion.dynserv.net/~timo/