On Wed, 2016-06-01 at 12:47 +0200, Radek Vykydal wrote:
On 1.6.2016 09:16, Vratislav Podzimek wrote:
>
> On Wed, 2016-06-01 at 08:51 +0200, Jiří Konečný wrote:
> >
> > On Tue, 2016-05-31 at 14:35 +0200, Radek Vykydal wrote:
> > >
> > > + network_data = self.assert_parse("network --device
> > > eth0 --
> > > activate --no-activate")
> > > + self.assertEquals(network_data.activate, False)
> > > + network_data = self.assert_parse("network --device
> > > eth0 --
> > > no-activate --activate")
> > > + self.assertEquals(network_data.activate, True)
> > Shouldn't this two tests give us an error/exception instead?
> > I see this as a big place to generate errors from missing one of
> > these
> > parameters in a command.
> >
> > I think good behavior here should be to let user know that these
> > commands can't be used together.
> AFAICT, it's basically impossible to check for this now with both
> options using
> the 'activate' attribute. And the policy is that the last one wins
> which should
> be okay and many command line tools work the same way with such
> options.
>
This is exactly my opinion.
I'm not quite convinced that this behavior is the good one but if it's
impossible to check then you have ACK from me too.