On Tue, 5 May 2009, Mark J. Nelson wrote:
> This note was blind copied to on-discuss at opensolaris.org. If you read
> it there and want to follow the conversation, please subscribe to
> Now that we've got 400+ contributions under our collective belt, I
> imagine that everybody's got opinions about what's good, bad, and ugly
> about the whole request/sponsor process.
> And we want to hear what we can do to streamline that process, making it
> easier for Contributors and for Sponsors.
> Here's one idea that's been percolating for a little while now, targeted
> primarily at more experienced Contributors:
> Write an optional Mercurial extension to (A) collect, through a series
> of questions and answers, supplemented by changeset comments, all of the
> information necessary for a sponsor to file an RTI; and (B) submit that
That seems very useful to me.
> information as the explanatory text for an "hg email" command , where
> a sponsor with commit rights can pick it up and oversee the mechanics of
> integration. The separation of items (A) and (B) is important--(A) is
> intended as a stopgap measure until we get the resources to figure out
> how OpenRTI fits, and (B) is a first step towards changing the way we
> integrate changes.
> For a complex fix, and/or a relatively inexperienced Contributor, this
> won't look much different from what we have today. A sponsor will still
> be available via the request-sponsor alias, and can still help with any
> part of the internal process. The earlier the sponsor is involved, the
> less useful this extension would be.
> For more experienced submitters, however, this would have the following
> - Allow greater autonomy: a note to request-sponsor will still be
> appropriate to avoid duplicating effort, but now it becomes a
> notification that "I'm working on this," instead of a request for a sponsor.
With how our processes are currently set up, a sponsor (or someone
internal) will still need to then update the CR internally noting
that it is being worked on. Not all developers for ON are on the
request-sponsor alias - in fact, just a small handful are.
> - Provide a more formal means of handing off the fix: right now, fixes
> may be communicated in a variety of formats. This will unify and
> formalize the expected format, which is a necessary prerequisite for
> automating processes such as rebase, recommit, and validation. Such
> automation is explicitly not part of this phase.
This would be nice. Presumably, this would put more of the onus of
testing, etc, on the external contributor?
> - Allow more flexibility: the more information that the Contributor
> provides, the less interactive the handoff may be. The less information
> provided, the more interactive. The less experienced the Contributor,
> the earlier a Sponsor is assigned, and the more the Sponsor participates
> in the process.
> For now, I would love to hear from Contributors and Sponsors. Does this
> sound like a step in a useful direction? Would you use such a thing?
yes, though I don't think it will solve all our woes, it will help.
Thank you for thinking about this,
Valerie Fenwick, http://blogs.sun.com/bubbva
Solaris Security Technologies, Developer, Sun Microsystems, Inc.
17 Network Circle, Menlo Park, CA, 94025.