From: Mark Levedahl <mlevedahl@gmail.com>
Date: Tue, 14 Oct 2014 01:05:47 GMT
Subject: Re: [PATCH] remote.c - Make remote definition require a url
Message-ID: <543C76EB.1000103@gmail.com>
In-Reply-To: <xmqqppdv51lv.fsf@gitster.dls.corp.google.com>

On 10/13/2014 01:19 PM, Junio C Hamano wrote:
> Mark Levedahl <mlevedahl@gmail.com> writes:
>
>> Some options may be configured globally for a remote (e.g, tagopt).
> Or some remotes may have only pushurl and not url.  "git remote"
> output for me has a few such remotes but wouldn't this patch break
> it?
>
> If a caller that walks the list of remotes misbehaves only because
> it assumes that r->url always is always valid, isn't that assumption
> what needs to be fixed?  for_each_remote() should be kept as a way
> to enumerate all the [remote "foo"], I would think.
>
>
>
>
As long as the rule is that for_each_remote will enumerate every remote 
that has anything defined at all, even if only in the global config 
outside of a user's control, I'm not really sure how to tell whether the 
missing url / pushurl / whatever is intentional, or a misconfiguration, 
so having the code complain that it didn't find what it wanted (the 
current condition) is probably no worse than the alternatives. Patch 
withdrawn.

Mark