From: Jonathan Gordon <jdgordy_at_gmail.com>
Date: Sun, 10 Oct 2010 16:05:29 +1100

On 9 October 2010 23:39, Daniel Stenberg <daniel_at_haxx.se> wrote:
> On Sat, 9 Oct 2010, Jonathan Gordon wrote:
>
>> We really need to have a proper discussion on how to get
>> consensus/agreement from the (active) dev group.
>
> My suggestion for what's required to get a new feature added:
>
> Three devs (expressed) in favour (+1), and none being against (-1) given
> enough time to react (ie more than 24 hours).
>
> If there's anyone agaist it, there must be reasons specified for the
> negative vote and there should be a discussion for what's needed to make the
> veto go away.
>
> I believe this is roughly how the Apache foundation does it. If we want to,
> we can check that up more careful to see if we can adapt their "rules" even
> more verbatim.
>
> Just my idea.
>
> --
>
>  / daniel.haxx.se
>

I have a few problems with this idea.
First, what is the bar to decide if we actually need to vote before a
commit? everything? everything not trivial?
Secondly, ignoring opposition like "This could be coded better" why
should one person have veto power? Especially when quite a few times
the only opposition comes from people who don't see the value of an
addition (and so wouldn't affect them anyway)?

Also, there is a big difference between rockbox and Apache. Apart from
user base and funding, anything contributed to the Apache foundations
various projects could be huge attack vectors so security is a big
deal, fortunately this isnt such an issue for us.

Jonathan


P.S Extra bureaucracy is bad! I really think the only thing that needs
changing is peoples attitudes and reactions to something they aren't
entirely happy with. Everything is version controlled so a civil
discussion post commit is just as valid as precommit. (with the added
benefit of more people actually trying out said change before shooting
it down)
Received on 2010-10-10