[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20200625141359.GL442307@lunn.ch>
Date: Thu, 25 Jun 2020 16:13:59 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Stephen Hemminger <stephen@...workplumber.org>
Cc: Michal Kubecek <mkubecek@...e.cz>, netdev <netdev@...r.kernel.org>,
Chris Healy <cphealy@...il.com>
Subject: Re: [ethtool v2 4/6] Add --json command line argument parsing
On Wed, Jun 24, 2020 at 10:32:58PM -0700, Stephen Hemminger wrote:
> On Thu, 25 Jun 2020 02:12:42 +0200
> Andrew Lunn <andrew@...n.ch> wrote:
>
> > diff --git a/internal.h b/internal.h
> > index edb07bd..7135140 100644
> > --- a/internal.h
> > +++ b/internal.h
> > @@ -23,6 +23,8 @@
> > #include <sys/ioctl.h>
> > #include <net/if.h>
> >
> > +#include "json_writer.h"
> > +
> > #define maybe_unused __attribute__((__unused__))
> >
> > /* internal for netlink interface */
> > @@ -221,6 +223,8 @@ struct cmd_context {
> > int argc; /* number of arguments to the sub-command */
> > char **argp; /* arguments to the sub-command */
> > unsigned long debug; /* debugging mask */
> > + bool json; /* Output JSON, if supported */
> > + json_writer_t *jw; /* JSON writer instance */
>
> You can avoid the boolean by just checking for NULL jw variable.
Hi Stephen
It is a while since i wrote this code, but i think i considered
that. The problem is, only a few commands support json output. I could
call json_new() unconditional of if the command actually support json
or not, that is not a problem. But then json_destory() should also be
unconditionally called. And that does fputs("\n", self->out); So you
end up with an extra blank line.
Using the boolean allows me to defer json_new()/json_destroy() into
the actual commands which supports json.
Andrew
Powered by blists - more mailing lists