[Gnu-arch-users] Re: Linus

[Miles Bader]

"Stephen J. Turnbull" <address@hidden> writes:
>     Miles> Using the above scheme, everything seems to work, at the
>     Miles> cost of two special groups for the emacs project -- which
>     Miles> doesn't seem a very high price to pay, and also seems like
>     Miles> an extremely natural way to set things up.
> 
> At XEmacs, we enable commits to a single Lisp package for anybody who
> is an upstream package maintainer, or has the agreement of the
> upstream maintainer to maintain the XEmacs package.  At present there
> are about 35 people with restricted commit privileges, and I would say
> that all of them are different, so we'd need nearly 40 groups.

Sure, but you certainly can't do something that sophisticated without a
special arch server (note that you want to restrict based on the _source_
file, which requires groveling through the changeset to validate in arch).

My point was that those things an arch `dumb server' _can_ do, needn't
require much from the server administrator or changes to arch.  I mean, I
don't oppose every change to arch for access control purposes, but my
impression is that the changes being pushed in this thread are a bit
questionable.

-Miles
-- 
Fast, small, soon; pick any 2.