You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by Benson Margulies <bi...@gmail.com> on 2012/04/27 15:10:24 UTC
Maven
The seemingly endless debate about Maven decamped from this list to a
JIRA, and continued to be apparently endless. However, it seems to me
that it really isn't all that far away from a reasonable resolution.
The following is an attempt to summarize the decision points.
1: Should the Lucene PMC publish maven artifacts at all?
- There seems to remain a consensus of the community: "Yes".
2: Should the publication be a seamless aspect of the release process,
or a secondary step after the coast is clear? (And, if secondary,
should snapshots be pushed automatically?)
3: Should the Lucene PMC ever take responsibility for distributing
other people's stuff?
This has been one of the more fraught questions, and needs to be
divided up a bit.
3a: Should the PMC formally vote and release source of modified
versions of other people's stuff, and in particular other Apache TLP's
stuff? Keeping in mind strong guidance from the board to the effect
that packages must be renamed in this case.
3b: Should the PMC include modified binaries that it creates in 'lib'
directories or as Maven artifacts? (That is, advertise them for
inclusion in users' application classpaths) Keeping in mind strong
guidance from the board to the effect that packages must be renamed in
this case. Also note that pulling in a binary from github or maven
central is, I claim, noncontroversial.
3c: Should the PMC include modified binaries in binary release
packages without advertising them for inclusion in users' application
classpaths? Renaming is not required here, but the source release has
to be buildable.
- I think that the consensus here is 'yes'.
4: If the answer to some pieces of (3) is a 'no', should individuals
of the Lucene PMC take steps to arrange for the convenient existence
of releases of modified versions of other people's stuff?
In thinking about item 4, I invite commentators to distinguish 'a
bugfix that fixes a problem for some people some of the time' from 'a
change that is absolutely necessary for the Lucene PMC release to work
at all'.
5: Are there enough Maven-maven contributors to ensure that
Maven-related issues are dealt with swiftly?
6: (Semi-seriously) are there enough Apache policy geek contributors
to ensure that releases comply with policy, on all fronts ranging from
the minutia of NOTICE files to the questions above under (3)?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org
RE: Maven
Posted by Steven A Rowe <sa...@syr.edu>.
Bill,
ASF policy is that discussions that don't need to be private should not be.
Steve
-----Original Message-----
From: Bill Bell [mailto:billnbell@gmail.com]
Sent: Friday, April 27, 2012 9:42 PM
To: dev@lucene.apache.org
Cc: dev@lucene.apache.org
Subject: Re: Maven
Why are you debating this here? Isn't there a PMC place to discuss topics with the main PMC members?
Bill Bell
Sent from mobile
On Apr 27, 2012, at 7:24 PM, Steven A Rowe <sa...@syr.edu> wrote:
> Hi Benson,
>
> On 4/27/2012 at 8:36 PM, Benson Margulies wrote:
>> On Fri, Apr 27, 2012 at 2:27 PM, Steven A Rowe <sa...@syr.edu> wrote:
>>> On 4/27/2012 at 9:10 AM, Benson Margulies wrote:
>>>> 1: Should the Lucene PMC publish maven artifacts at all?
>>>>
>>>> - There seems to remain a consensus of the community: "Yes".
>>>
>>> Not really: consensus means unanimous agreement.
>>
>> No, consensus means no one chooses to block consensus. And since
>> Maven is in place, it would take a decision by consensus to remove
>> it.
>
> According to <http://www.apache.org/foundation/voting.html>, procedural votes require majority approval, not concensus.
>
> But maybe you're referring to the code change vote procedure, which
> allows a single PMC member to veto code changes? IMHO, Maven support
> in Lucene/Solr doesn't qualify as code. (Simple test: does Maven
> support - the POMs - ship with the source releases? The answer is
> no.)
>
>> Much email has been sent on the premise that Maven specially and
>> uniquely raises certain IP and policy issues. It doesn't.
>> It does offer some constraints and opportunities for coping with
>> these issues.
>
> AFAICT, Lucene/Solr completely complies with ASF policies, and will continue to do so - this is a requirement for the continued existence of the project under ASF governance.
>
> By contrast, Maven's "offered constraints", including its separate publishing platform, present challenges that need to be addressed separately, at a cost that the project is not required to bear. The nature of those challenges (IP/policy/whatever) is IMHO immaterial.
>
>> However, if I'm not being helpful, and it seems that I'm not, I'll return to lurking.
>
> I value your opinions even though I may not always share them. And I certainly welcome your assistance in meeting the challenges of keeping Maven in-house here.
>
> Steve
>
> Т
> ХF
> V 7V'67& &R R âFWb V 7V'67& &T V6V R 6 R &pФf "FF F 6
> G2 R âFWbֆV V6V R 6 R &pР
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org For additional commands, e-mail: dev-help@lucene.apache.org
Re: Maven
Posted by Bill Bell <bi...@gmail.com>.
Why are you debating this here? Isn't there a PMC place to discuss topics with the main PMC members?
Bill Bell
Sent from mobile
On Apr 27, 2012, at 7:24 PM, Steven A Rowe <sa...@syr.edu> wrote:
> Hi Benson,
>
> On 4/27/2012 at 8:36 PM, Benson Margulies wrote:
>> On Fri, Apr 27, 2012 at 2:27 PM, Steven A Rowe <sa...@syr.edu> wrote:
>>> On 4/27/2012 at 9:10 AM, Benson Margulies wrote:
>>>> 1: Should the Lucene PMC publish maven artifacts at all?
>>>>
>>>> - There seems to remain a consensus of the community: "Yes".
>>>
>>> Not really: consensus means unanimous agreement.
>>
>> No, consensus means no one chooses to block consensus. And since
>> Maven is in place, it would take a decision by consensus to remove
>> it.
>
> According to <http://www.apache.org/foundation/voting.html>, procedural votes require majority approval, not concensus.
>
> But maybe you're referring to the code change vote procedure, which allows a single PMC member to veto code changes? IMHO, Maven support in Lucene/Solr doesn't qualify as code. (Simple test: does Maven support - the POMs - ship with the source releases? The answer is no.)
>
>> Much email has been sent on the premise that Maven specially
>> and uniquely raises certain IP and policy issues. It doesn't.
>> It does offer some constraints and opportunities for coping
>> with these issues.
>
> AFAICT, Lucene/Solr completely complies with ASF policies, and will continue to do so - this is a requirement for the continued existence of the project under ASF governance.
>
> By contrast, Maven's "offered constraints", including its separate publishing platform, present challenges that need to be addressed separately, at a cost that the project is not required to bear. The nature of those challenges (IP/policy/whatever) is IMHO immaterial.
>
>> However, if I'm not being helpful, and it seems that I'm not, I'll return to lurking.
>
> I value your opinions even though I may not always share them. And I certainly welcome your assistance in meeting the challenges of keeping Maven in-house here.
>
> Steve
>
> Т���������������������������������������������������������������������ХF�V�7V'67&�&R�R���âFWb�V�7V'67&�&T�V6V�R�6�R��&pФf�"FF�F����6����G2�R���âFWbֆV��V6V�R�6�R��&pР�
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org
Re: Maven
Posted by Benson Margulies <bi...@gmail.com>.
Steven,
I was afraid that my message would come out cranky, I really didn't
mean it that way.
I entirely endorse your view that 3.6.0 was completely legitimate. My
harping on policy is my last gasp at convincing Rob and others that
the unhappy interaction with the larger foundation over commons-csv
was only incidentally a result of the use of Maven. Either I'm wrong
about that or I'm hopelessly unpersuasive, so I'll stop spilling
electrons on the subject.
--benson
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org
RE: Maven
Posted by Steven A Rowe <sa...@syr.edu>.
Hi Benson,
On 4/27/2012 at 8:36 PM, Benson Margulies wrote:
> On Fri, Apr 27, 2012 at 2:27 PM, Steven A Rowe <sa...@syr.edu> wrote:
>> On 4/27/2012 at 9:10 AM, Benson Margulies wrote:
>>> 1: Should the Lucene PMC publish maven artifacts at all?
>>>
>>> - There seems to remain a consensus of the community: "Yes".
>>
>> Not really: consensus means unanimous agreement.
>
> No, consensus means no one chooses to block consensus. And since
> Maven is in place, it would take a decision by consensus to remove
> it.
According to <http://www.apache.org/foundation/voting.html>, procedural votes require majority approval, not concensus.
But maybe you're referring to the code change vote procedure, which allows a single PMC member to veto code changes? IMHO, Maven support in Lucene/Solr doesn't qualify as code. (Simple test: does Maven support - the POMs - ship with the source releases? The answer is no.)
> Much email has been sent on the premise that Maven specially
> and uniquely raises certain IP and policy issues. It doesn't.
> It does offer some constraints and opportunities for coping
> with these issues.
AFAICT, Lucene/Solr completely complies with ASF policies, and will continue to do so - this is a requirement for the continued existence of the project under ASF governance.
By contrast, Maven's "offered constraints", including its separate publishing platform, present challenges that need to be addressed separately, at a cost that the project is not required to bear. The nature of those challenges (IP/policy/whatever) is IMHO immaterial.
> However, if I'm not being helpful, and it seems that I'm not, I'll return to lurking.
I value your opinions even though I may not always share them. And I certainly welcome your assistance in meeting the challenges of keeping Maven in-house here.
Steve
Re: Maven
Posted by Benson Margulies <bi...@gmail.com>.
On Fri, Apr 27, 2012 at 2:27 PM, Steven A Rowe <sa...@syr.edu> wrote:
> On 4/27/2012 at 9:10 AM, Benson Margulies wrote:
>> 1: Should the Lucene PMC publish maven artifacts at all?
>>
>> - There seems to remain a consensus of the community: "Yes".
>
> Not really: consensus means unanimous agreement.
No, consensus means no one chooses to block consensus. And since Maven
is in place, it would take a decision by consensus to remove it.
However, at the end of the day, that's for you all to decide.
>
> The rest of your points, Benson, confuse Maven issues around 3rd party artifacts with project governance issues; I won't address them here.
I'm trying to remove confusion. Much email has been sent on the
premise that Maven specially and uniquely raises certain IP and policy
issues. It doesn't. It does offer some constraints and opportunities
for coping with these issues.
However, if I'm not being helpful, and it seems that I'm not, I'll
return to lurking.
--benson
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@lucene.apache.org
For additional commands, e-mail: dev-help@lucene.apache.org
RE: Maven
Posted by Steven A Rowe <sa...@syr.edu>.
On 4/27/2012 at 9:10 AM, Benson Margulies wrote:
> 1: Should the Lucene PMC publish maven artifacts at all?
>
> - There seems to remain a consensus of the community: "Yes".
Not really: consensus means unanimous agreement.
The rest of your points, Benson, confuse Maven issues around 3rd party artifacts with project governance issues; I won't address them here.
I think action needs to be taken to address Maven's impact on the project.
Up till now, the bargain was something like: Maven proponents will take care of Maven issues, and others won't have to care. But the 3.6 release cleanup (thanks very much to Robert Muir, Chris Male, Uwe Schindler and others who did the hard work) stands as an example of what will be continuing issues. Robert's summary on SOLR-3405 is on point:
> We still dont have:
>
> - a way to handle patched dependencies for maven
> - a way to handle dependencies that are not in maven
> - a packaging system for maven consistent with our other packaging.
I want to get Lucene/Solr Maven stuff to the point where non-Maven-proponents don't have anything substantive to complain about. In the past I've worked toward addressing people's concerns about Maven in Lucene/Solr. I plan on continuing to do so.
Steve