You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@river.apache.org by Mark Brouwer <ma...@cheiron.org> on 2007/06/03 11:32:13 UTC
When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept the
Starter Kit Contribution from Sun Microsystems]
I started a new thread for I was under the impression it was policy not
to have discussions on a vote thread as it complicates tallying.
I agree with the recollection of John. It would be good if those with a
more in depth knowledge of the rules here could explain the
possibilities with regard to releasing in the com.sun.jini namespace.
The comparison with Wicket came up, and although I don't think it
was the intention of Gianugo to compare River with Wicket I believe
Wicket is a complete different project. They did just one release after
which they directly asked for graduation to TLP which only has to be
formalized by the board I believe. They came in Incubator after us and
are almost out of the gate, they had a diverse community to start with
that only needed to sort out the legal matters and to learn "the Apache
way". Although my memory is faulty and vague, I also believe they did or
intended to do bug-fix releases at Codehaus while doing their work here
in Incubation.
It is my expectation that River will be much longer in Incubation due to
the fact this was a completely Sun driven effort in terms of coding.
Although estimations are hard to give I won't be surprised as it would
be in line with a project such as Derby.
For me moving to org.apache.river in the first releases (if these have
to be soon) while having no sight of a graduation is something I
consider a bridge too far. Not only do I have a lot of patches in the
com.sun.jini namespace that needs to be integrated, I also have a lot of
derivative work based on the com.sun.jini namespace that is in
production at places and that needs ongoing support. For me switching is
something I really would like to postpone until the larger outlines for
this project are clear, e.g. are we going to make incompatible API
changes in the com.sun.jini API, etc.
Personally I have no problem with no official release in com.sun.jini
and let people build one from SVN themselves [1] and postpone an
official release in the org.apache.river namespace if we have some signs
we are heading the right way, but I believe others do.
[1] I realize this comes to the expense of user satisfaction, but I also
believe this way they become a little bit more involved and get a bit
more understanding of the project. As I believe incubation is also meant
for building a developer community I don't consider this 'bad'. A proper
explanation near the download section will do fine I guess.
--
Mark
Re: When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept
the Starter Kit Contribution from Sun Microsystems]
Posted by Mark Brouwer <ma...@cheiron.org>.
Craig L Russell wrote:
> Just as a thought-provoker, how about October 2007 for the namespace
> change?
Without pinning myself to it, my gut feeling tells me this is doable.
--
Mark
Re: When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept the
Starter Kit Contribution from Sun Microsystems]
Posted by Craig L Russell <Cr...@Sun.COM>.
Please forgive the top post, I just have a couple of observations.
There seems to be a conflict between the desire to get the community
on board at Apache and to create a new community by making a quick
release. The first group want to use the com.sun.jini namespace to
get their stuff working in the new environment; the second group need
to use the org.apache.river namespace.
It might be that you just have to decide on a time frame during which
you continue to use the com.sun namespace, have a project that
builds, tests, and basically works. During this time, anyone who
wants to get their stuff working can do so with minimum disruption.
But you don't make a release of these bits. Everyone is a source
developer.
Later, when stuff actually works, you can as a community decide on
the disruptive change of namespace and then consider how to make the
code available to a wider audience. I agree with Mark that picking
the time frame is probably the hardest thing to do.
Just as a thought-provoker, how about October 2007 for the namespace
change?
Craig
On Jun 4, 2007, at 2:40 AM, Mark Brouwer wrote:
> Gianugo Rabellino wrote:
>
>> I see your point, and I understand how backward compatibility
>> matters.
>> On the other hand, my main issue is about branding, which has to deal
>> with diversity as well. More specifically:
>> a) from a branding perspective, it should be clear to users that this
>> is Apache River, not Sun's Jini (or possibly the Apache
>> implementation of Sun's Jini specs). This is a clear Incubator
>> requirement.
>> b) I believe that diversity builds mainly upon neutrality: it's
>> harder
>> to find people willing to work on com.sun.* (think competitors), as
>> they'd feel they'll be working on Sun stuff, not ASF's.
>> c) while I can see your point in changing packages as jumping the
>> cliff, sometimes you just have to burn your boats. Yes, River might
>> fail incubation, which would mean a mess to the Jini community. On
>> the
>> other hand, playing safe with package names might be a sign of
>> missing
>> commitment to the community.
>
> I agree and understand your points. But besides that I would like
> to be
> able to transfer bells and whistles while it is easy to do so, I'm
> also
> of the opinion that any disruptive change for the Jini community
> should
> be something that is well thought over and be tested in the field by
> some early adopters. To me this seems hard to unite with a quick first
> release as has been expressed by some people.
>
> FWIW ome background information for those with a short history with
> Jini. From the Davis project (2002) I can remember that Sun
> developed a
> lot of then future specs in the com.sun.jini namespace because it was
> not a standard at that time (i.e. they couldn't use the net.jini
> namespace) and from what I can remember the namespace refactoring
> was a
> significant one with a lot of impact for early adopters. IIRC there
> were
> discussions in the Technical Oversight Committee about using the
> net.jini namespace at an earlier stage to prevent from this happening.
>
> While I think jumping cliff is necessary at some point, when near a
> shore I don't want to have a burnt boat with me. I care a lot of the
> current installed base and those who invested in the Jini technology
> since a long time, which includes the selfish me :-). I want them
> to be
> able to start using and getting confidence in the work of the Apache
> River project and I think a good way to achieve that is when they are
> able to replace their current binaries and see their systems are still
> running fine.
>
> If we can achieve that confidence we will get to the point of doing
> disruptive changes (such as the namespace changes, maybe going the
> a new
> minimal version of Java, etc.). That move will bring costs to those
> with
> an installed base, but at least they might justify that by seeing or
> even participating in a project that has at least has shown it can act
> as a coherent bunch of individuals.
>
>> I see your point, and I fundamentally agree. All I'm asking is that
>> package renaming sits somewhere high in the priority list.
>
> I think based on discussions in the past we all see changing name
> spaces
> as a logical consequence of coming to the ASF.
>
> Probably the only disagreement will be with regard to the time at
> which
> we should conduct such an effort (hopefully at that time I can take a
> long holiday and we can find one person we supply plenty of beer
> who is
> going to do the job (I think it ain't something that allows for a
> lot of
> parallelism, but no doubt one of the Sun people can tell from previous
> experience what it involves). It will also be a good moment to sort
> out
> our whitespace issues (this one is for you Gregg :-P ), etc.
> --
> Mark
>
Craig Russell
Architect, Sun Java Enterprise System http://java.sun.com/products/jdo
408 276-5638 mailto:Craig.Russell@sun.com
P.S. A good JDO? O, Gasp!
Re: When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept
the Starter Kit Contribution from Sun Microsystems]
Posted by Gregg Wonderly <gr...@wonderly.org>.
Mark Brouwer wrote:
> Probably the only disagreement will be with regard to the time at which
> we should conduct such an effort (hopefully at that time I can take a
> long holiday and we can find one person we supply plenty of beer who is
> going to do the job (I think it ain't something that allows for a lot of
> parallelism, but no doubt one of the Sun people can tell from previous
> experience what it involves).
The only thing that I'd like to say about the repackaging issue is that we might
just consider using subclassing and/or proxy implementations of the same class
names in org.apache as the initial thing to do, so that both namespaces are
visible. Putting deprecated notices on the com.sun.jini stuff would then be the
indicator that new code shouldn't be using it. It might also be worth changing
the packaging on the existing code to org.apache and using subclassing/proxy
implementations to recreate the com.sun.jini classes. I just feel that there
would be less potential impact to do the "org.apache is proxy impl" first so
that less code relationships change and the new namespace is available. Later,
we can do the swap so that org.apache is the real code with com.sun.jini
proxying to it.
> It will also be a good moment to sort out
> our whitespace issues (this one is for you Gregg :-P ), etc.
I have shared my opinion about this, and apparently many people started
programming after disk space was no longer a concern so that the reduction in
whitespace file space consumption was not a problem :-) I appreciate tabs
because then I can set tab stops to whatever I like and the file is
appropriately structured. Others appreciate whitespace so that they see the
file the way everyone else sees it. Still others don't know what they
appreciate because they've always done it how their editor/IDE environment
introduces these things. Let's just move on and decide on something that will
unify it, and I'll figure out how to deal with it (probably a perforce branch).
Gregg Wonderly
Re: When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept
the Starter Kit Contribution from Sun Microsystems]
Posted by Mark Brouwer <ma...@cheiron.org>.
Gianugo Rabellino wrote:
> I see your point, and I understand how backward compatibility matters.
> On the other hand, my main issue is about branding, which has to deal
> with diversity as well. More specifically:
>
> a) from a branding perspective, it should be clear to users that this
> is Apache River, not Sun's Jini (or possibly the Apache
> implementation of Sun's Jini specs). This is a clear Incubator
> requirement.
>
> b) I believe that diversity builds mainly upon neutrality: it's harder
> to find people willing to work on com.sun.* (think competitors), as
> they'd feel they'll be working on Sun stuff, not ASF's.
>
> c) while I can see your point in changing packages as jumping the
> cliff, sometimes you just have to burn your boats. Yes, River might
> fail incubation, which would mean a mess to the Jini community. On the
> other hand, playing safe with package names might be a sign of missing
> commitment to the community.
I agree and understand your points. But besides that I would like to be
able to transfer bells and whistles while it is easy to do so, I'm also
of the opinion that any disruptive change for the Jini community should
be something that is well thought over and be tested in the field by
some early adopters. To me this seems hard to unite with a quick first
release as has been expressed by some people.
FWIW ome background information for those with a short history with
Jini. From the Davis project (2002) I can remember that Sun developed a
lot of then future specs in the com.sun.jini namespace because it was
not a standard at that time (i.e. they couldn't use the net.jini
namespace) and from what I can remember the namespace refactoring was a
significant one with a lot of impact for early adopters. IIRC there were
discussions in the Technical Oversight Committee about using the
net.jini namespace at an earlier stage to prevent from this happening.
While I think jumping cliff is necessary at some point, when near a
shore I don't want to have a burnt boat with me. I care a lot of the
current installed base and those who invested in the Jini technology
since a long time, which includes the selfish me :-). I want them to be
able to start using and getting confidence in the work of the Apache
River project and I think a good way to achieve that is when they are
able to replace their current binaries and see their systems are still
running fine.
If we can achieve that confidence we will get to the point of doing
disruptive changes (such as the namespace changes, maybe going the a new
minimal version of Java, etc.). That move will bring costs to those with
an installed base, but at least they might justify that by seeing or
even participating in a project that has at least has shown it can act
as a coherent bunch of individuals.
> I see your point, and I fundamentally agree. All I'm asking is that
> package renaming sits somewhere high in the priority list.
I think based on discussions in the past we all see changing name spaces
as a logical consequence of coming to the ASF.
Probably the only disagreement will be with regard to the time at which
we should conduct such an effort (hopefully at that time I can take a
long holiday and we can find one person we supply plenty of beer who is
going to do the job (I think it ain't something that allows for a lot of
parallelism, but no doubt one of the Sun people can tell from previous
experience what it involves). It will also be a good moment to sort out
our whitespace issues (this one is for you Gregg :-P ), etc.
--
Mark
Re: When com.sun.jini -> org.apache.river [Was: Re: [VOTE] Accept the Starter Kit Contribution from Sun Microsystems]
Posted by Gianugo Rabellino <gi...@apache.org>.
On 6/3/07, Mark Brouwer <ma...@cheiron.org> wrote:
> I started a new thread for I was under the impression it was policy not
> to have discussions on a vote thread as it complicates tallying.
Absolutely, thanks for this.
> I agree with the recollection of John. It would be good if those with a
> more in depth knowledge of the rules here could explain the
> possibilities with regard to releasing in the com.sun.jini namespace.
Best thing to do is bringing this on the general incubator list once
we have a set of clear questions to ask and one or two possibilities
ahead.
> The comparison with Wicket came up, and although I don't think it
> was the intention of Gianugo to compare River with Wicket I believe
> Wicket is a complete different project.
Absolutely agreed. On the other hand, consider how they were all in
all namespace neutral (they were using wicket.*).
> It is my expectation that River will be much longer in Incubation due to
> the fact this was a completely Sun driven effort in terms of coding.
> Although estimations are hard to give I won't be surprised as it would
> be in line with a project such as Derby.
Agreed, that's likely to happen.
> For me moving to org.apache.river in the first releases (if these have
> to be soon) while having no sight of a graduation is something I
> consider a bridge too far.
I see your point, and I understand how backward compatibility matters.
On the other hand, my main issue is about branding, which has to deal
with diversity as well. More specifically:
a) from a branding perspective, it should be clear to users that this
is Apache River, not Sun's Jini (or possibly the Apache
implementation of Sun's Jini specs). This is a clear Incubator
requirement.
b) I believe that diversity builds mainly upon neutrality: it's harder
to find people willing to work on com.sun.* (think competitors), as
they'd feel they'll be working on Sun stuff, not ASF's.
c) while I can see your point in changing packages as jumping the
cliff, sometimes you just have to burn your boats. Yes, River might
fail incubation, which would mean a mess to the Jini community. On the
other hand, playing safe with package names might be a sign of missing
commitment to the community.
> Not only do I have a lot of patches in the
> com.sun.jini namespace that needs to be integrated,
I assume this can be easily done once code lands in SVN?
> I also have a lot of
> derivative work based on the com.sun.jini namespace that is in
> production at places and that needs ongoing support.
Good point, but there is a clear trade-off with branding and community
building. I'm sure there is a sweet spot somewhere which might be
satisfactory. Maybe a compatibility package (no idea whether it's
feasible at all) that bridges/proxies com.sun.* to org.apache.*? Or
some other idea we could come up with.
> For me switching is
> something I really would like to postpone until the larger outlines for
> this project are clear, e.g. are we going to make incompatible API
> changes in the com.sun.jini API, etc.
I see your point, and I fundamentally agree. All I'm asking is that
package renaming sits somewhere high in the priority list.
> Personally I have no problem with no official release in com.sun.jini
> and let people build one from SVN themselves [1] and postpone an
> official release in the org.apache.river namespace if we have some signs
> we are heading the right way, but I believe others do.
>
> [1] I realize this comes to the expense of user satisfaction, but I also
> believe this way they become a little bit more involved and get a bit
> more understanding of the project. As I believe incubation is also meant
> for building a developer community I don't consider this 'bad'. A proper
> explanation near the download section will do fine I guess.
Yes, that could be an option.
Ciao,
--
Gianugo Rabellino
Sourcesense, making sense of Open Source: http://www.sourcesense.com
Orixo, the XML business alliance: http://www.orixo.com
(blogging at http://www.rabellino.it/blog/)