FOSS and Interface Taxonomy levels

[Nicolas Williams]

On Wed, Mar 05, 2008 at 10:23:49AM -1000, Joseph Kowalski wrote:
> Observation: Having a well architected repository for well triaged
> components would be less expensive than having this discussion again
> and again and again.

We're stepping up to do the FOSS integration.  Every time someone does
who hasn't seen this discussion, a replay ensues.  That's an education
issue, is it not?

> Observation: Delivery of "nethack" is important enough that even it
> can't wait for a repository.  Well, I never would have thought....

If we refrain from having this discussion and educate the integrator
then we can probably make progress faster.

Time for an FAQ that can be linked to in response to anyone beginning
a replay of this thread?

Now, if there really, truly is a need to come to a "Volatile is good for
more things" conclusion then deferring that by knocking integrator heads
(to get them to go with Uncommitted or better) will cause a waste of
effort.  But I don't believe it.

The mail utilities case was instructive.  Some parts were made Volatile
and others Uncommitted and all are happy.  The same applies to much of
the browser stack, no?  Sure, the mail utilities case did not have
libraries, but we have cases that are representative of the extremes of
the spectrum when it comes to libraries (SQLite3 and OpenSSL).

What more do we need to settle this discussion for good (for a while)?

Nico
--