[Opensim-dev] The notion of "core"

[Ryan McDougall]

On Tue, Oct 20, 2009 at 2:49 AM, Frisby, Adam <adam at deepthink.com.au> wrote:
> Ter pretty much summed it up - both it and the irc channel are fairly low-volume, and the 'topic' is restricted to only 'personal' or 'meta' matters; such as discussion of approval of commit rights.
>
> It's pretty standard practice across open source projects with more than 5 committers for the committers to have a mailing list for these purposes, since realtime chats aren't practical across timezones.
>
> Adam
>

I am not sure I'd agree just how standard a process it is.

The one's I've been involved with or otherwise have some detailed
knowledge of, have never had them; including such big names as GNOME,
Fedora, and Linux. For example the GNOME foundation list is not only
world-readable, but anyone can join:
http://mail.gnome.org/mailman/listinfo/foundation-list . Actual
foundation members are voted by the community at large.

Basically the way they are able to operate is, they don't distribute
commit access according to monolithic vote of knighted members; they
have a system of maintainership, and each maintainer gives access
rights to his module/repo as she sees fit, in a web of trust.

One of the complaints one sometimes hears is how monolithic the
project is (even if the code-base is modular). Maybe the move to git,
and the maturation of the code allows more distribution and
specialization of responsibility?

My concerns with core mailing list are:

1. It's "secret", ie. not world readable. I can understand limiting
membership to voting partners to avoid bikeshedding, but I can't
understand secrecy of any kind in an open source project.

2. Decisions made there (aside from commit rights) affect other
people, and they not only have no voice to represent themselves, they
don't even get to know what is being said about them. That doesn't
seem fair somehow.

The knowledge that someone can read what you write makes you think
harder about what you say. Maybe a private list makes the problem of
disagreement within core worse rather than better? I haven't the
faintest idea who this snowcrash guy is, but when I was a topic of
discussion on -core, I remember not liking it at all.

As for the issue of timezones, I understand that completely! Which is
why I wish you guys used ML more frequently! :)

My intention is not to bike-shed, but to be productive. Either opensim
core is open to this point of view or it's not, and we move on from
there.

Cheers, and much love!