[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Policy about policy

On Mon, Sep 06, 1999 at 02:01:18PM +1000, Anthony Towns wrote:
> They're not, after all, all *that* bad. Sure, we had a whole bunch of
> problems with /usr/doc, amongst which were:
> 	* 'twas a complicated issue with no elegant solutions
> 	* everyone didn't really know how to work with `formal objections'
> 	* no one really knew how to work with the technical committee (IMHO
> 	  the -policy group looked on it as an easy way out, then a month
> 	  or so later realised they didn't have their act together either)
> 	* we'd made a major change in policy that made no explicit consessions 
> 	  to backwards compatability

This seems reasonably accurate.

I think, however, that "no elegant solutions" should be a warning flag
that we're doing something wrong.

> We've come up with solutions (or at least countermeasures) to most
> of these. First, `formal objections' definitely should be able to
> be responded to, and withdrawn. That didn't really happen with
> this proposal, and if it had, I suspect we'd have managed to get a
> consensus anyway, eventually. Second, the technical committee at least
> has a working mailing list and a chair now. We even, eventually, got a
> response.

I think Wichert gets credit for getting the technical committee
up and running.

> Finally, we are at least considering making changes that say "the
> current way is okay, but we'd prefer it if you did such-n-such", which
> IMO would make changes like this one much easier to bear (witness the
> `-g' proposal, for example. Ideally it wants major changes to every
> package, but it doesn't complain about unmodified packages being non
> policy-compliant).

Yes, I agree that this is a good thing.

> So I don't see a huge need for upheaval. I think the policy groups
> (-policy and the tech ctte) are heading in the right direction.


I don't think I see a need for a huge upheaval either (someone was
suggesting that there wouldn't be much point for the policy list, the
more I think about it, the more I disagree with that point of view).

I do think that the policy group needs to pay a bit more attention
to their own resolutions/procedures/informal practices/whatever you
call them.  I do think that the policy group should consider actively
soliciting the opinions of package developers [on the basis that 
package developers have some significant expertise in a variety of
matters].  It is possible that some documents (proposal, constitution,
etc.) need some fairly subtle edits.

But huge upheaval?  That's exactly what I'm trying to avoid.

> > Personally, I'd hate to rewrite the proposal document without any
> > input from the policy maintainers, and likewise, I'd hate to propose
> > a constitutional change to match the current proposal document
> > without even more input [enough to see that most developers are
> > unhappy with the idea that technical decisions are better made by
> > individuals than by large groups].
> That's not quite the way -policy works though: it's not a "big groups"
> versus "individuals" thing, it's a consensus thing. I'm not going to
> attempt to define it more precisely than that: I'm sure I'd get it
> horribly wrong and I think we've already got good enough means for
> testing consensus anyway, and a definition would merely get in the
> way.

I think I understand, and I think I have a nit: consensus is a fine
thing, but you have to be careful to distinguish between apathy and
consensus.  [Which is yet another reason to expect that soliciting
the opinions of developers would be a good thing.]

> In particular, the policy group *does* decide on technical matters
> by consensus, whether those decisions conflict with other packages
> or not. Yes, suggesting major changes instead of demanding them is
> /better/, and I'm sure we'll take that into account in future, but
> it's only really important in one of a dozen (</pretend_statistics>)
> cases: we shouldn't redo the whole of the -policy mechanism about it.

Yeah, sample size is too small for statistics to be very meaningful.

However, an 8% failure rate isn't really acceptable for major policy
decisions -- in my opinion.  [Not unless you really do expect another
group to comb through what you've approved for quality control purposes.]

> What I'd like to see is the technical committee and the DPL ratify the
> powers of the -policy group, rather than try to impose changes from
> above.

Please review section 3 of the constitution for an overview of the
[constitutional] powers of policy maintainers.

Like you, the constitution doesn't really define consensus -- but I think
that the informal definition "as long as there's no formal objection"
is completely inadequate.


Reply to: