Re: [nft PATCH v2] libnftables: Unexport enum nftables_exit_codes

[Pablo Neira Ayuso <pablo@netfilter.org>]

On Mon, Nov 13, 2017 at 02:49:04PM +0100, Phil Sutter wrote:
> On Mon, Nov 13, 2017 at 01:31:00PM +0100, Pablo Neira Ayuso wrote:
> > On Fri, Nov 10, 2017 at 12:27:15PM +0100, Phil Sutter wrote:
> > > diff --git a/src/main.c b/src/main.c
> > > index 529bedffc2e3b..8d03f8989b1fc 100644
> > > --- a/src/main.c
> > > +++ b/src/main.c
> > > @@ -183,11 +183,11 @@ int main(int argc, char * const *argv)
> > >  		switch (val) {
> > >  		case OPT_HELP:
> > >  			show_help(argv[0]);
> > > -			exit(NFT_EXIT_SUCCESS);
> > > +			exit(0);
> > 
> > Better use the standard EXIT_FAILURE and EXIT_SUCCESS here? Instead of
> > hardcoded 0 and 1 values.
> 
> While at it, should I convert nft_run_cmd_from_*() to return those
> macros as well? This would of course mean changing failure case from -1
> to 1, but would streamline calls from nft.c. What do you think?

I prefer to keep EXIT_* in main() only.

To me, -1 is the standard way to return an error in functions. So
that's fine to me to keep it around.

So my suggestion is, you focus on features and fixes at this stage,
cleanups may come later on, so we avoid too much code churning, ie.
things that we may need to undo later on when extending things.

Cleanups can follow up once dust settles down a bit.

Let me know, thanks.