You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jcp-open@apache.org by Dain Sundstrom <da...@iq80.com> on 2006/03/01 03:53:36 UTC
Re: JSR 291 : OSGi
I find this JSR very confusing. Why is the package name
org.osgi.framework? Is this just a rubber stamp for a spec not
developed within the JCP?
-dain
On Feb 28, 2006, at 10:44 AM, Geir Magnusson Jr wrote:
> Today a new JSR was officially announced :
>
> http://www.jcp.org/en/jsr/detail?id=291
>
> As you can see, this is for OSGi.
>
> I've told IBM that we support the JSR, and will participate in the
> Expert Group.
>
> Please volunteer if you are interested in representing the ASF on
> this EG.
>
> I've CC-ed felix-dev. Please don't crosspost, but discuss on jcp-
> open@apache.org
>
> geir
Re: JSR 291 : OSGi
Posted by Geir Magnusson Jr <ge...@pobox.com>.
Alan D. Cabrera wrote:
> Geir Magnusson Jr wrote, On 3/1/2006 9:27 AM:
>
>>
>>
>> Dain Sundstrom wrote:
>>
>>> I'm curious if this means that the OSGi spec committee outside of the
>>> JCP will shutdown. Otherwise I expect we will be rubber stamping
>>> everyone's outside specs.
>>
>>
>> The answer to the org.osgi question was that yes, the idea is to
>> preserve the existing namespace, similar to org.omg...
>>
>> And I don't think that OSGi will shutdown. I think that the purpose
>> here is to have the JCP produce a statement regarding how OSGi is used
>> in a standard way.
>>
>>>
>>> I am concerned about any JSR or any path to a JSR that subverts the
>>> community process.
>>
>>
>> I'm not sure what is subverted here. There are a bunch of things,
>> such as webservices, for which the JCP does a similar thing - namely
>> define the official API to work with the technology...
>
>
> Your analogy does not hold. The difference is that OSGi _already_ has a
> Java API spec. WS-* was in need of Java APIs, hence the need for JCP
> participation.
Yes, I realize the difference. Hoever, a difference is that unlike WS*,
there is a two-way relationship here - because if a JSR did JCP-ize the
OSGi interface, it at least should modify the control point for the
technology to where the JCP community has a say in future evolution.
>
> I do not understand where the value add is in bringing in the
> participation of the JCP.
>
Well, it's a good question. The real answer will come from IBM, of course.
The only value I see is that the JCP is a group we know and participate
in, and the OSGi Federation isn't, and we have an apache community
(felix) that is interested in this, as they actually are implementing OSGi.
geir
Re: JSR 291 : OSGi
Posted by "Alan D. Cabrera" <ad...@toolazydogs.com>.
Geir Magnusson Jr wrote, On 3/1/2006 9:27 AM:
>
>
> Dain Sundstrom wrote:
>
>> I'm curious if this means that the OSGi spec committee outside of the
>> JCP will shutdown. Otherwise I expect we will be rubber stamping
>> everyone's outside specs.
>
>
> The answer to the org.osgi question was that yes, the idea is to
> preserve the existing namespace, similar to org.omg...
>
> And I don't think that OSGi will shutdown. I think that the purpose
> here is to have the JCP produce a statement regarding how OSGi is used
> in a standard way.
>
>>
>> I am concerned about any JSR or any path to a JSR that subverts the
>> community process.
>
>
> I'm not sure what is subverted here. There are a bunch of things,
> such as webservices, for which the JCP does a similar thing - namely
> define the official API to work with the technology...
Your analogy does not hold. The difference is that OSGi _already_ has a
Java API spec. WS-* was in need of Java APIs, hence the need for JCP
participation.
I do not understand where the value add is in bringing in the
participation of the JCP.
Regards,
Alan
Re: JSR 291 : OSGi
Posted by Dain Sundstrom <da...@iq80.com>.
On Mar 1, 2006, at 11:12 AM, Geir Magnusson Jr wrote:
> Dain Sundstrom wrote:
>> On Mar 1, 2006, at 9:27 AM, Geir Magnusson Jr wrote:
>>> Dain Sundstrom wrote:
>>>> I'm curious if this means that the OSGi spec committee outside
>>>> of the JCP will shutdown. Otherwise I expect we will be rubber
>>>> stamping everyone's outside specs.
>>>
>>> The answer to the org.osgi question was that yes, the idea is to
>>> preserve the existing namespace, similar to org.omg...
>>>
>>> And I don't think that OSGi will shutdown. I think that the
>>> purpose here is to have the JCP produce a statement regarding how
>>> OSGi is used in a standard way.
>>>
>>>> I am concerned about any JSR or any path to a JSR that subverts
>>>> the community process.
>>>
>>> I'm not sure what is subverted here. There are a bunch of
>>> things, such as webservices, for which the JCP does a similar
>>> thing - namely define the official API to work with the
>>> technology...
>>>
>>> Given the current state of the community process in the JCP, how
>>> do you see this subverting it?
>> As the US Supreme Court says "I know it when I see it". FWIU,
>> there are only two places where the JCP deferred to an external
>> group CORBA and WebServices. In both cases, these are well
>> established standards bodies, and they the specs are cross
>> language. Again FWIU, OSGi is purely Java focused and was not an
>> established specification group when it started.
>
> Nor is it now, is it?
No offense to the OSGi people, but I would say no. If you walk up to
the average Java developer and ask them if they have heard of the JCP
or W3C, I would expect them to say yep. If you asked about the OSGi
Alliance, I doubt it would be the same level.
>>> Are there other models our EG reps would want to propose?
>> In the case of OSGi, I see no reason for the JCP to map any
>> statement about OSGi. It is outside of the community process. If
>> they want to join the community process. Then I would expect to
>> see the OSGi Alliance to shutdown and we start working on OSGi r5
>> within the JCP.
>
> That would be cool.
>
>> Alternatively, if the JCP is now willing to rubberstamp
>> specifications developed outside of the JCP,
>
> I do think that the rubberstamping has happened before, and I don't
> think that there is any general problem with it. For example,
> that's what the Groovy strategy seems to be - develop outside of
> the JCP's restrictive IP rules in an OSS project, and then throw it
> over the fence when done for rubberstamping. It's not a perfect
> analogy though.
Really? I personally think of Groovy as how specs should work. You
discuss everything publicly, develop the RI and TCK publicly all
under the EG. Then when everything looks good, you put it to the EC
for a vote. Other than the word "publicly" isn't that how all EGs work?
> I believe that there are far more examples of rubber stamping in
> the ME space, but I'm not very conversant in those JSRs.
I think that is a problem. Where is the community in a rubber
stamping process? It just seems counter to open source and I'm not
sure we want to support such specs.
>> I suggest that Apache investigate forming a new spec foundation
>> outside of the JCP that better represents our values. We can
>> then develop specifications with others and bring them
>> to the JCP for rubber stamping. I'm sure this is what everyone
>> else will do now that it is allowed ;)
>
> This has been suggested before, and is always an option :)
>
> I keep hoping someone invents a better, radically different
> approach to standardization then the ways we do it now, in Java and
> elsewhere.
I think what we have in the JCP is workable, but I would like to see
more meritocracy in the EGs. I think this change will make things
worse not better.
-dain
Re: JSR 291 : OSGi
Posted by Geir Magnusson Jr <ge...@pobox.com>.
Dain Sundstrom wrote:
> On Mar 1, 2006, at 9:27 AM, Geir Magnusson Jr wrote:
>
>> Dain Sundstrom wrote:
>>> I'm curious if this means that the OSGi spec committee outside of the
>>> JCP will shutdown. Otherwise I expect we will be rubber stamping
>>> everyone's outside specs.
>>
>> The answer to the org.osgi question was that yes, the idea is to
>> preserve the existing namespace, similar to org.omg...
>>
>> And I don't think that OSGi will shutdown. I think that the purpose
>> here is to have the JCP produce a statement regarding how OSGi is used
>> in a standard way.
>>
>>> I am concerned about any JSR or any path to a JSR that subverts the
>>> community process.
>>
>> I'm not sure what is subverted here. There are a bunch of things,
>> such as webservices, for which the JCP does a similar thing - namely
>> define the official API to work with the technology...
>>
>> Given the current state of the community process in the JCP, how do
>> you see this subverting it?
>
> As the US Supreme Court says "I know it when I see it". FWIU, there are
> only two places where the JCP deferred to an external group CORBA and
> WebServices. In both cases, these are well established standards
> bodies, and they the specs are cross language. Again FWIU, OSGi is
> purely Java focused and was not an established specification group when
> it started.
Nor is it now, is it?
>
>> Are there other models our EG reps would want to propose?
>
> In the case of OSGi, I see no reason for the JCP to map any statement
> about OSGi. It is outside of the community process. If they want to
> join the community process. Then I would expect to see the OSGi
> Alliance to shutdown and we start working on OSGi r5 within the JCP.
That would be cool.
>
> Alternatively, if the JCP is now willing to rubberstamp specifications
> developed outside of the JCP,
I do think that the rubberstamping has happened before, and I don't
think that there is any general problem with it. For example, that's
what the Groovy strategy seems to be - develop outside of the JCP's
restrictive IP rules in an OSS project, and then throw it over the fence
when done for rubberstamping. It's not a perfect analogy though.
I believe that there are far more examples of rubber stamping in the ME
space, but I'm not very conversant in those JSRs.
> I suggest that Apache investigate forming
> a new spec foundation outside of the JCP that better represents our
> values. We can then develop specifications with others and bring them
> to the JCP for rubber stamping. I'm sure this is what everyone else
> will do now that it is allowed ;)
This has been suggested before, and is always an option :)
I keep hoping someone invents a better, radically different approach to
standardization then the ways we do it now, in Java and elsewhere.
geir
Re: JSR 291 : OSGi
Posted by Dain Sundstrom <da...@iq80.com>.
On Mar 1, 2006, at 9:27 AM, Geir Magnusson Jr wrote:
> Dain Sundstrom wrote:
>> I'm curious if this means that the OSGi spec committee outside of
>> the JCP will shutdown. Otherwise I expect we will be rubber
>> stamping everyone's outside specs.
>
> The answer to the org.osgi question was that yes, the idea is to
> preserve the existing namespace, similar to org.omg...
>
> And I don't think that OSGi will shutdown. I think that the
> purpose here is to have the JCP produce a statement regarding how
> OSGi is used in a standard way.
>
>> I am concerned about any JSR or any path to a JSR that subverts
>> the community process.
>
> I'm not sure what is subverted here. There are a bunch of things,
> such as webservices, for which the JCP does a similar thing -
> namely define the official API to work with the technology...
>
> Given the current state of the community process in the JCP, how do
> you see this subverting it?
As the US Supreme Court says "I know it when I see it". FWIU, there
are only two places where the JCP deferred to an external group CORBA
and WebServices. In both cases, these are well established standards
bodies, and they the specs are cross language. Again FWIU, OSGi is
purely Java focused and was not an established specification group
when it started.
> Are there other models our EG reps would want to propose?
In the case of OSGi, I see no reason for the JCP to map any statement
about OSGi. It is outside of the community process. If they want to
join the community process. Then I would expect to see the OSGi
Alliance to shutdown and we start working on OSGi r5 within the JCP.
Alternatively, if the JCP is now willing to rubberstamp
specifications developed outside of the JCP, I suggest that Apache
investigate forming a new spec foundation outside of the JCP that
better represents our values. We can then develop specifications
with others and bring them to the JCP for rubber stamping. I'm sure
this is what everyone else will do now that it is allowed ;)
-dain
Re: JSR 291 : OSGi
Posted by Geir Magnusson Jr <ge...@pobox.com>.
Dain Sundstrom wrote:
> I'm curious if this means that the OSGi spec committee outside of the
> JCP will shutdown. Otherwise I expect we will be rubber stamping
> everyone's outside specs.
The answer to the org.osgi question was that yes, the idea is to
preserve the existing namespace, similar to org.omg...
And I don't think that OSGi will shutdown. I think that the purpose
here is to have the JCP produce a statement regarding how OSGi is used
in a standard way.
>
> I am concerned about any JSR or any path to a JSR that subverts the
> community process.
I'm not sure what is subverted here. There are a bunch of things, such
as webservices, for which the JCP does a similar thing - namely define
the official API to work with the technology...
Given the current state of the community process in the JCP, how do you
see this subverting it?
Are there other models our EG reps would want to propose?
geir
>
> -dam
>
> On Mar 1, 2006, at 2:52 AM, Geir Magnusson Jr wrote:
>
>>
>>
>> Dain Sundstrom wrote:
>>> I find this JSR very confusing. Why is the package name
>>> org.osgi.framework?
>>
>> Reasonable question... I'll ask. We'll probably hear that it's
>> either an oversight, or more - more likely - there to "preserve
>> compatibilty" with existing users of OSGi.
>>
>>> Is this just a rubber stamp for a spec not developed within the JCP?
>>
>> I don't know. It very well could be. I can imagine ways that this
>> could go that would be useful - such as devise a way to do
>> componentization w/ a runtime lifecycle for which OSGi is one
>> implementation - and ways which it would be "less useful" - like
>> rubberstamping an external spec.
>>
>> Now that I just re-read the JSR, looking at the proposed timeline, I
>> think I'll take "rubberstamp for $400", since I can't imagine how much
>> new technical work will get done in the 4 month timeline they are
>> proposing for the whole JSR.
>>
>> I'll ask about that too....
>>
>> geir
>>
>>
>>> -dain
>>> On Feb 28, 2006, at 10:44 AM, Geir Magnusson Jr wrote:
>>>> Today a new JSR was officially announced :
>>>>
>>>> http://www.jcp.org/en/jsr/detail?id=291
>>>>
>>>> As you can see, this is for OSGi.
>>>>
>>>> I've told IBM that we support the JSR, and will participate in the
>>>> Expert Group.
>>>>
>>>> Please volunteer if you are interested in representing the ASF on
>>>> this EG.
>>>>
>>>> I've CC-ed felix-dev. Please don't crosspost, but discuss on
>>>> jcp-open@apache.org
>>>>
>>>> geir
>
>
Re: JSR 291 : OSGi
Posted by Dain Sundstrom <da...@iq80.com>.
I'm curious if this means that the OSGi spec committee outside of the
JCP will shutdown. Otherwise I expect we will be rubber stamping
everyone's outside specs.
I am concerned about any JSR or any path to a JSR that subverts the
community process.
-dam
On Mar 1, 2006, at 2:52 AM, Geir Magnusson Jr wrote:
>
>
> Dain Sundstrom wrote:
>> I find this JSR very confusing. Why is the package name
>> org.osgi.framework?
>
> Reasonable question... I'll ask. We'll probably hear that it's
> either an oversight, or more - more likely - there to "preserve
> compatibilty" with existing users of OSGi.
>
>> Is this just a rubber stamp for a spec not developed within the JCP?
>
> I don't know. It very well could be. I can imagine ways that this
> could go that would be useful - such as devise a way to do
> componentization w/ a runtime lifecycle for which OSGi is one
> implementation - and ways which it would be "less useful" - like
> rubberstamping an external spec.
>
> Now that I just re-read the JSR, looking at the proposed timeline,
> I think I'll take "rubberstamp for $400", since I can't imagine how
> much new technical work will get done in the 4 month timeline they
> are proposing for the whole JSR.
>
> I'll ask about that too....
>
> geir
>
>
>> -dain
>> On Feb 28, 2006, at 10:44 AM, Geir Magnusson Jr wrote:
>>> Today a new JSR was officially announced :
>>>
>>> http://www.jcp.org/en/jsr/detail?id=291
>>>
>>> As you can see, this is for OSGi.
>>>
>>> I've told IBM that we support the JSR, and will participate in
>>> the Expert Group.
>>>
>>> Please volunteer if you are interested in representing the ASF on
>>> this EG.
>>>
>>> I've CC-ed felix-dev. Please don't crosspost, but discuss on jcp-
>>> open@apache.org
>>>
>>> geir
Re: JSR 291 : OSGi
Posted by Geir Magnusson Jr <ge...@pobox.com>.
Dain Sundstrom wrote:
> I find this JSR very confusing. Why is the package name
> org.osgi.framework?
Reasonable question... I'll ask. We'll probably hear that it's either
an oversight, or more - more likely - there to "preserve compatibilty"
with existing users of OSGi.
> Is this just a rubber stamp for a spec not
> developed within the JCP?
I don't know. It very well could be. I can imagine ways that this
could go that would be useful - such as devise a way to do
componentization w/ a runtime lifecycle for which OSGi is one
implementation - and ways which it would be "less useful" - like
rubberstamping an external spec.
Now that I just re-read the JSR, looking at the proposed timeline, I
think I'll take "rubberstamp for $400", since I can't imagine how much
new technical work will get done in the 4 month timeline they are
proposing for the whole JSR.
I'll ask about that too....
geir
>
> -dain
>
> On Feb 28, 2006, at 10:44 AM, Geir Magnusson Jr wrote:
>
>> Today a new JSR was officially announced :
>>
>> http://www.jcp.org/en/jsr/detail?id=291
>>
>> As you can see, this is for OSGi.
>>
>> I've told IBM that we support the JSR, and will participate in the
>> Expert Group.
>>
>> Please volunteer if you are interested in representing the ASF on this
>> EG.
>>
>> I've CC-ed felix-dev. Please don't crosspost, but discuss on
>> jcp-open@apache.org
>>
>> geir
>
>