Fwd: [Incubator] Project Incubation Mentor

Steve Lime Steve.Lime at dnr.state.mn.us
Fri Apr 21 19:24:35 EDT 2006


Since mapserver can easily be considered a toolkit I'd say you still have 15 minutes still to go... ;-)

Steve

>>> cholmes at openplans.org 04/21/06 5:05 PM >>>
Hrmmm...  This email I mostly composed while out in california, but 
never got around to sending, and I'm clearing out my drafts box.  Sorry 
if any of it's outdated.  I read the IRC I missed, and thanks for 
electing me 'mentor' for MapServer, it's certainly an honor, as it's 
always been the type of project my communities have aspired to.

...

Frank Warmerdam wrote:
> Sean Gillies wrote:
> 
>>> My perception is that the committee wants to fast track the initial 
>>> projects. That's what I mean by mainstream. IMO, the process should 
>>> be slower, more conservative, and have a real effect on the projects.
> 
> 
> Sean,
> 
> It is certainly true that I'm hoping for relatively rapid progress
> through incubation and have been pushing.  But pushing back is OK
> if there are effects we are looking for in incubation and we aren't
> achieving them in the short term.
 From a geotools perspective, I agree with Jody that we're more excited
about almost a 'slower' process.  We don't feel our act is quite enough
together with really accessible docs for new users.  I don't think we
want to call _ourselves_ an 'osgeo project' until we can be happier
about a user's initial experience.

And I agree that all project should have a minimal set of requirements
to pass the 'new user' test, on at least windows and linux.  I think
Frank and I have been informally doing this to eachother for years, with
both of us failing.  I couldn't get MapServer to install, or rather, I
got it to install (with ms4w), and still wasn't sure what to _do_ to see
a map.  He got as far as installing but never did the appropriate
plug-ins for java that GeoServer needed.

Basically I feel we should have something like the '5 minute' rule for
applications, 5 minutes of install time and you'd better have something
to play with.

For toolkits it can be more like the 20 minute rule, the time it takes
for a new developer to get set up with the code and doing the most
common tasks they'd want to.


> 
>>> Conservative to me means that only projects with a bright and shiny 
>>> future should be released from the incubator.
> 
> 
> I think we can all agree that we don't want moribund projects in the
> foundation.  It reflects poorly on the foundation, and it distracts
> and confuses users who assume that foundation software must be good
> and well supported.
> 
> What is less clear to me is how bright and shiny a project needs to
> be to be part of the foundation.  Obviously, we would like to attract
> the best projects because they offer the foundation the most reflected
> glory, and because the foundation is likely to accomplish the most with
> them.
> 
> But I don't personally see a problem with having a foundation project
> that works well, is mature and well supported but perhaps lacks the
> growth and excitement of some other projects.  Even a project in
> gentile and benign decline wouldn't necessarily be a bad thing though
> I can't imagine being to excited about inviting such a project in
> unless it filled a particularly important role.
Yeah, I'm inclined to say that excluding a project just because it's
post-mature and isn't going to be doing anything new and exciting is a
bad idea.  As long as it's still easy to configure and install, isn't
falling hopeless out of date, then such projects likely add a whole lot
to the foundation.  They are the ones that drive traffic, and allow _us_
to point people to the new OSGeo projects that we want.

> 
>  >> To be specific, I think
> 
>>> MapServer's best days are behind it. I'm not talking code, here. I'm 
>>> talking about developers and business. Much of MapServer's 
>>> development is mercenary. We've usually seen this as a sign of vigor: 
>>> people are
>>> making money as MapServer developers. On the other hand, mercenary 
>>> developers can always be employed away to other projects. I expect to 
>>> see Autodesk undermine MapServer like this. It's not evil, it's just 
>>> business. Additionally, the MapServer Enterprise/Cheetah fiasco 
>>> showed me how ready some businesses are to swap MapServer out for the 
>>> next big thing. IMO, MapServer should demonstrate that its mojo is 
>>> healthy and increasing before it's out of the incubator.
> 
> 
> Hmm, I'm not sure I agree with all the above, though it is an interesting
> opinion.
Yeah, I think it needs to show that it's healthy, but not necessarily
increasing.  Overall it has to show that it'd be an overall win for the
foundation to have it be a part.

> 
>>> GDAL, on the other hand, clearly has a bright future. What I'd like 
>>> to see there is a real indicator that the project is achieving its 
>>> goal of transition from benevolent tyranny to community and meritocracy. 
> 
> 
> I think it would be helpful for us to write up what it is we expect
> of projects in this regard.  I have written odds and ends into some
> of the incubator documents about having a PSC that is in control, and
> is functioning well, but we haven't been to clear in our goals for
> responsible governance.
> 
> I feel that at a minimum there needs to be a PSC with at least 3
> members from different organizations.  I would like to know that these
> PSCs are open to outside input, and would be willing to add new
> qualified members as appropriate.  What I fear are closed cliques.
> I'm just not sure how to test this without trying to join. :-)
I think there are some decent ways to do this.  Like being able to find
in the docs how one achieves commit rights.  I recently realized we
didn't do this in GeoServer.  There has to be documented procedures for
you one can go from a patcher to a committeer to being on a PSC that can
make binding decisions based on majority.  Of course a project could
just write rules and not really follow them, but that's easier for us to
see, and just having them up goes a lont way.

> 
> And I want to know that the PSC is really in control.  That is
> that things aren't just run by a dictator with the PSC as a facade
> that is disregarded at will.  Actually, I'd also like to see that
> the PSC is control of things in other regards as well!  That random
> committers aren't just doing all sorts of crazy stuff without regard
> to contribution guidelines and due process.
Yes.

> 
> I'm not so sure what to make of "meritocracy".
> 
> I'd be interested in hearing what you (and others!) think is
> desirable and ought to be required in terms of community.  In my
> efforts so far I have emphasized that that there ought to be a
> support forum / mailing list where questions are reasonably
> likely to be answered.  To me this is one of the most important
> manifestations of community.  That users speak to users about
> how to solve problems.
Definitely.  Active lists, ideally something like irc as well where
people can get quicker answers.

> 
>>> What would the indicator be? I don't know for sure. Independent 
>>> mercenary developers or consultants might be a good sign. We always 
>>> thought they were for MapServer. (Note that it is very possible that 
>>> I am just unaware of existing independent GDAL developers and 
>>> consultants.) I think the same measure could be usefully applied to 
>>> single-shop projects like MapGuide OS.
> 
> 
> Hmm, I think you have a point about independent consultants in the case
> of GDAL.  There are other developers doing work as part of organizations
> (IDRISI, PCRaster, ITC, USGS, LANL, etc) and there are "independent"
> developers but the so called independent developers (such as Andrey
> Kiselev, Denis Nadeau, or Howard Butler) are often subcontracted by
> me which means they aren't exactly independent of me.  I certainly
> think it would be a sign of project health to have independent
> developer/consultants finding their own contracts for work that feeds
> back as improvements to GDAL.
> 
>>> I don't intend that going slow be punitive, but that it give adequate 
>>> time for projects to make progress on the transformations that 
>>> they've begun.
> 
> 
> I can agree that it takes some time to check that transitions to
> a working PSC can take time.  I'm not sure what other aspects of
> incubation require time to settle out.  I think the code copyright
> review is done when done.  Infrastructure transitions are done when
> they are done.
> 
> So, for a project like MapServer that already had a healthy "OSGeo
> style PSC" I think that it could be ready for approval once it
> passes the (IPish) hurdles.  Of course, that is assuming we agree
> that it's community and vitality is "good enough", but from my
> point of view this is not in doubt.
> 
> For projects like GDAL, OSSIM, GRASS, MapBender or MapGuide, I think
> we have  something to prove with regard to open and mature governance
> and this may take some time.  I think MapBuilder, and GeoTools are
> pretty mature in this regard, though I don't know them well enough
> to be sure.
> 
> But ultimately we, and the board on our advice, has to decide what
> we think is important in foundation projects, and develop mechanisms
> to test for it, and/or inculcate it.
> 
> Best regards,
> 

-- 
Chris Holmes
The Open Planning Project
thoughts at: http://cholmes.wordpress.com






More information about the Incubator mailing list