incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joe Schaefer <>
Subject Re: [PROPOSAL] Validation incubator for JSR-303 Bean Validation
Date Thu, 31 Dec 2009 04:03:40 GMT
----- Original Message ----

> From: Phil Steitz <>
> To:
> Sent: Wed, December 30, 2009 3:10:47 PM
> Subject: Re: [PROPOSAL] Validation incubator for JSR-303 Bean Validation
> Joe Schaefer wrote:
> > ----- Original Message ----
> > 
> >> From: Phil Steitz 
> >> To:
> >> Sent: Wed, December 30, 2009 1:30:13 PM
> >> Subject: Re: [PROPOSAL] Validation incubator for JSR-303 Bean Validation
> >>
> >> Joe Schaefer wrote:
> >>> ----- Original Message ----
> >>>
> >>>> From: ant elder 
> >>>> To:
> >>>> Sent: Fri, December 11, 2009 5:22:13 AM
> >>>> Subject: Re: [PROPOSAL] Validation incubator for JSR-303 Bean Validation
> >>>>
> >>>> On Fri, Dec 11, 2009 at 9:56 AM, Niall Pemberton
> >>>> wrote:
> >>>>> On Fri, Dec 11, 2009 at 7:56 AM, ant elder wrote:
> >>>>>> A quick search so there has been some discussion on commons-dev
- [1]
> >>>>>>
> >>>>>> Does this really need to be incubated - the proposal says its
> >>>>>> to graduate to Apache Commons and replace the existing Validator
> >>>>>> component as a new 2.0 codebase, from the discussion on commons-dev
> >>>>>> everyone seems fine with that out come, and only 2 of the 7
> >>>>>> committers are not existing Validator or ASF committers - so
> >>>>>> this just go straight to commons as a code grant and make the
two new
> >>>>>> guys committers in recognition of contibuting the new code?
> >>>>> I raised this on private@commons and reported back to dev@commons
> >>>>> that discussion here:
> >>>>>
> >>>>>
> >>>>>
> >>>>> In summary though, there was very little support to go that route
> >>>>> some objections.
> >>>>>
> >>>>> All commons components share the same set of mailing lists which
> >>>>> it easier for PMC members to provide oversight for the 30+ components
> >>>>> that live there. As part of this proposal we want to use the commons
> >>>>> mailing lists for commits and discussion so that by the time this
> >>>>> podling is ready to graduate the new committers and Commons PMC
> >>>>> have a better knowledge of each other and there will be no issue
> >>>>> voting in the new committers.
> >>>>>
> >>>>> The use of the commons mailing lists is in the proposal and was
> >>>>> of the vote held on dev@commons to sponsor this incubation effort:
> >>>>>
> >>>>>
> >>>>>
> >>>>> Niall
> >>>>>
> >>>> From the first email referenced was Roman ever asked if he'd mind
> >>>> submitting patches for a while to earn Karma if the code did go
> >>>> straight to commons? Seems a bit a of a shame to need to go the whole
> >>>> incubation process just for one commit access.
> >>>>
> >>>> Re the the poddling use the existing commons mailing lists its may be
> >>>> worth pointing out this recent thread:
> >>>>
> >>> Commons is badly busted if it can't allow a new person access to his/her
> >>> own code in a fucking sandbox.  Incubating this project because some weenies

> >> are
> >>> uncomfortable about the nature of the meritocracy over in commons isn't
> >> solution:
> >>> have commons hold a public vote and make an actual decision.  If they vote

> to
> >>> incubate the damned thing, it's an incredibly stupid decision, but so be
> >>>
> >> Hey Joe, the language could be toned down a bit, but I see your
> >> point.  On the other hand, here is the problem as I see it.
> >>
> >> In Commons, like other non-Incubator projects, we welcome new
> >> contributors and encourage them to get involved in the community and
> >> stick around long enough to earn ASF commit.  When people show up
> >> with significant patches, we ask them to file CLAs before we commit
> >> their code and if the contribution is "big" (not precisely defined,
> >> but we have been able to agree in all cases), we ask for a software
> >> grant and go through Incubator IP clearance.  We have several
> >> examples of people showing up with large amounts of code, engaging
> >> in the community and contributing patches to their own and other
> >> code and earning commit that way.  This has worked for us in the
> >> past and is consistent with how things are supposed to work - at
> >> least as I understand it - at the ASF, outside of the Incubator.  If
> >> we have changed our (ASF) view on what it means to become a
> >> committer, then maybe we are behind the times in Commons.  That
> >> would be somewhat ironic, since in the Jakarta days we were
> >> regularly accused of having too low a bar for commit.
> >>
> >> What we would have no problem at all with is following the process
> >> described above - just do IP clearance / code grant for the code and
> >> let the non-ASF committers earn commit.  This does not take forever
> >> and is not as terrible as some seem to think it is.  I can't recall
> >> a single "failure" (someone getting discouraged and giving up) and
> >> several successes over the past 6 years.
> >>
> >> I understand that in the Incubator people get commit immediately and
> >> that makes it easier for both them and the mentors.  As I understand
> >> it, part of the reason we have the Incubator is so that people who
> >> have no experience with the ASF and have not earned merit can both
> >> gain experience and demonstrate merit in a "mentored" environment.
> >> The mentoring and graduation requirements ensure that when projects
> >> graduate, their committers have earned full ASF commit.
> > 
> > I seriously doubt mentors take their role that seriously, otherwise
> > we wouldn't have so many long-term residents of the Incubator.
> > 
> >> It could be that I have this wrong and just arriving with a lump of
> >> code that a project wants to incorporate is enough to earn ASF
> >> commit outside the Incubator nowadays.  If we collectively agreed to
> >> that and I missed the conversation, then I apologize for the late
> >> protest.  I honestly can't believe that we did agree to that; however.
> >>
> >> Note that this has nothing to do with expectations about who will
> >> succeed, who will not - it is about meritocracy being based on
> >> publicly earned merit. Good code is good and if unencumbered we can
> >> commit it to our projects.  Good people interested in open
> >> development make good committers and these we can include in our ASF
> >> committer community. It is our collective responsibility to give as
> >> many people as possible the opportunity to succeed in becoming
> >> committers; but they have to do more than just produce good code to
> >> earn commit.
> >>
> >> If we want to extend the Incubator function to work within projects,
> >> I am OK with that. That could be the right way to deal with
> >> situations such as this - though in this particular case, I still
> >> think just software grant/earn commit is best - and it could take
> >> some load off of the Incubator PMC. The role of the "sponsoring PMC"
> >> would then become in loco Incubator. I am open to this, but need to
> >> understand better how exactly it would work.
> >>
> >> Phil
> > 
> > I'm not looking to extend the Incubator even in the smallest way.  To me
> > the Incubator is a necessary evil, and not entirely a positive thing for
> > the ASF.  That it seems to be functional on some level is good enough for
> > the time being- at least we're back to graduating projects again.
> > 
> > Where I disagree with your outlook is in the relative meaning of granting
> > someone commit.  To me it's just giving someone the full set of tools and
> > the right (conditional on peer review) to make changes to a codebase.  Each
> > community can and should make their own determination on what additional
> > meaning they wish to impart to committership, but communities need to be
> > flexible, not rigid, in how they make those determinations.
> So should commit require merit or not?  I understand the need for
> flexibility, but if we are going to have a meaningful "committer"
> role, we need to be consistent (at least within projects) what it
> means.  At commons, we are not rigid, but we do require sustained
> contributions (vaguely defined, but at least a few months) before we
> vote someone in as a committer.  This is because being voted in
> means something in terms of merit.  We could change that and divorce
> commit from merit - postponing the merit requirement until the PMC
> membership vote.  Is this what you mean?

I don't see any need for several months to pass before being able to figure
out if someone is participating to further the group's goals and understands
the group's unique style of work.  That is the distinguishing characteristic
of whether someone should be given full access to subversion IMO, not some
nebulous notion of accrued merit.  The point of having a version control system
in place is that we can be lax in how we dish out permissions to it *because*
it's easy to fix mistakes *after* they happen.  The overriding goal is to weed
out people who consume more collective energy than they give back, not to bestow
an honorific title on those that clear the bar.

The reason we give out commit access is so people can be full participants in
collaborative work at the ASF.  There are additional benefits when it comes to
conference fees, but that's the gist of it.  Everything else is artificial.

As Niall points out, there seems to be some open question as to how much interest
there is within the commons community to provide support for this effort.  That's
not made any easier by moving things into the Incubator IMO.  Personally if I were
Niall and were still willing to mentor these people, it would make my life easier
if they had commit to the sandbox so I could comment on their full range of work
habits.  If you wanted to you could lock down their commit access to just the sandbox.
OTOH it'd be a big PITA to proxy their commits by having them submit patches for
me to process.  In other words, be lazy and optimize for the mentor's time, because
that is the scarce resource here.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message