incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nicola Ken Barozzi <>
Subject Re: Clearing the air round Incubator and Ruper (was Re: projects incubated by the incubator PMC)
Date Thu, 11 Dec 2003 07:34:05 GMT

Sander Striker wrote:

> On Wed, 2003-12-10 at 18:49, Brian Behlendorf wrote:
>>On Wed, 10 Dec 2003, Nicola Ken Barozzi wrote:
>>>First of all, we are in the process of deciding and clearly documenting
>>>that only TLPs are to be incubated. Why? Because in Apache there are
>>>only TLPs. Thus, Ruper is incubated on the premises that it wants to
>>>become a TLP for artifact handling.
>>Are you "in the process of deciding", or is this a decision?  Where is the
>>decision being made?

I'm happy you asked :-)

There was a discussion on the thread '[RT] Incubator Reorg', and then I 
posted a proposal as '[PROPOSAL] Incubator Reorg', to which only Sam 

>>I don't feel strongly pro or con, but if this is the case, then don't be
>>surprised if the amount of confusion about "when should some non-trivial
>>amount of code coming into the ASF go through the incubator?" increases.
>>Many of the edge cases I've seen discussed (such as Wagon) have been
>>natural extensions of or adjuncts to products released by an existing TLP.
> There was a discussion on the board list at the end of 2002.  What was
> basically the point was that every external codebase would come through
> Incubator.  No exceptions (for obvious reasons).  IIRC the conclusion
> was that no PMC was to accept new projects on their own anymore.

Maven is in the process of adding Wagon and other codebases, and they 
are not passing by the Incubator. Hence in my proposal all PMC chairs 
are to be in the Incubator automatically to be able to do IP clearance.

> There is ofcourse always the question between when a trivial patch
> becomes non-trivial enough to warrant going through incubator. I
> personally think that when you have to ask, you've got a incubator
> candidate.

Seems good enough for me, as checking things that other PMCs are doing 
is IIUC only a board prerogative.

>>Often it'll even be a proposal for a next-dot-oh of an existing release.
>>Or another module for Apache httpd written by a company interested in
>>donating it to the httpd project for inclusion into the core or as part of
>>a contrib-modules release by the httpd project.  So this could have a
>>chilling effect on existing TLPs when confronted with the question,
>>"has this contribution surpassed a threshold and should be sent through
>>the incubator?".
> The reason to get Incubator involved is to be sure the paperwork is
> in order.  If a new mod_whatever is to be donated we need a software
> grant, for instance.  To not have to put the burden of knowing what
> paperwork is needed on each TLP, this knowledge is bundled in Incubator
> (at least, that's the idea).

Then we have a process and policy issue, as we currently don't cater for 
this scenario. @see the '[PROPOSAL] Incubator Reorg' thread.

>>I don't think you're going to see the board mandate that each TLP have
>>only one product.  A single "community" can easily manage/oversee the
>>release of multiple products.
> Even semingly single project TLPs have multiple projects.  HTTP Server
> has httpd, httpd-test, flood, apreq, just to name a few.  I agree that
> we don't want every project to be a TLP.  But maybe Nicola meant that
> from a legal standpoint there are only TLPs.

I mean that the legal and actual *responsibility* of the well being of a 
community (not product) is of TLPs, which are the only ones legally 

I'm talking of recognized communities here, not products. A community 
can make more than one product.

In essence, I could say that all *products* entering Apache have to pass 
by the Incubator, but only *communities* that want to go top level must 
enter the Incubator. If the PMC requires community incubation, we may as 
well help them, but it's not compulsory.

>> First of all, we are in the process of deciding and clearly
>> documenting that only TLPs are to be incubated. Why? Because in
>> Apache there are only TLPs.
> I don't agree. The Incubator has a dual role. One is the building of
> new Projects, but the other is vetting all incoming codebases. Some
> confusion comes from applying the term "incubation" to both aspects
> of its role.

Exactly. Since currently we only incubate communities, I use this 
meaning here.

> Please review the thread in the Board archives that I pointed out to 
> Jason. As Sander noted as well, it was clearly stated that the
> Incubator is responsible for all incoming codebases developed outside
> of the Foundation. No, patches don't need to come through the
> incubator, and yes, there are grey areas, but the clearly stated
> intent was that the Board had designated the Incubator as the sole
> place through which an externally developed codebase could enter the
> ASF.
> Where the confusion comes from is related to the Community building 
> usage of the term "incubation." The other PMCs have expressed that
> they want to do the Community Building when a codebase and a
> community is merging into their larger community. That is fine, but
> it does not change the Board's mandate regarding oversight of
> externally developed code. If the incoming code has a designated
> place to go, we should facilitate the PMC being able to start their
> Community Building, both we are responsible for ensuring that the IP
> issues are cleared, adn they are responsible for working with us to
> make sure it happens.

This is the idea.

Nicola Ken Barozzi         
             - verba volant, scripta manent -
    (discussions get forgotten, just code remains)

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

View raw message