Re: [PATCH] remote: filter out invalid remote configurations

From: Carlos Martín Nieto <cmn@elego.de>
Date: Aug 30, 2013, 12:41 UTC
Message-ID: <1377866509.1714.0.camel@centaur.cmartin.tk>
In-Reply-To: <xmqqr4dffarq.fsf@gitster.dls.corp.google.com>

On Tue, 2013-08-27 at 07:50 -0700, Junio C Hamano wrote:

Show 16 quoted lines

> Carlos Martín Nieto <cmn@elego.de> writes:
> 
> > In remote's configuration callback, anything that looks like
> > 'remote.<name>.*' creates a remote '<name>'. This remote may not end
> > up having any configuration for a remote, but it's still in the list,
> > so 'git remote' shows it, which means something like
> >
> >     [remote "bogus"]
> >         hocus = pocus
> >
> > will show a remote 'bogus' in the listing, even though it won't work
> > as a remote name for either git-fetch or git-push.
> 
> Isn't this something the user may want to be aware of, though?
> Hiding these would rob a chance for such an entry to be noticed from
> the user---is it a good change?

If we want to help the user know that there's something a bit odd in their configuration, shouldn't we tell them instead of hoping they stumble upon it? Otherwise IMO it's more confusing if git-remote does show the remote when git-fetch is interpreting the argument as a path.

cmn