On 1/18/17 08:55 , Guillaume Nodet wrote:
2017-01-18 14:29 GMT+01:00 Richard S. Hall <he...@ungoverned.org>:
On 1/18/17 08:22 , Guillaume Nodet wrote:
2017-01-18 13:53 GMT+01:00 Carsten Ziegeler <cziege...@apache.org>:
Whoever is doing the RI
does it somewhere else and might do a code contribution or not.
Yes, that definitely would avoid the problem.
And I don't think it changes anything from the contributor point of view :
the reason is that it's not really developed openly, as I explained, so
there's definitely no difference with donating the code once the spec has
been released.
There is no difference? Really? Claiming the current approach is not
optimal from a community perspective is certainly not unreasonable, but
saying that the community doesn't benefit at all from having draft
implementations being worked on at Apache seems like a stretch.
Can you outline the benefits then ? Honestly, I don't see the difference
between such an implementation being developed at Apache and the same one
being developed under the same license at github. At github, I can also
fork, provide PR, raise issues, etc...
Well, at a minimum we get a clearer and simpler IP trail. But at least
for community members who want to contribute or be exposed to what is
going on, we don't have to tell them "go someplace else and figure it
out". So, yes, honestly, I do see a difference.
Not sure
if that is the preferable way. We might end up with not having an
implementation at Apache at all.
The ASF does not care if there is one, multiple or no implementation at
Apache at all afaik. However, it cares about the way the community work
and that it operates as a meritocracy, which definitely rules out the fact
not all members have access to the same information.
This seems like splitting hairs to me. Even with a complete spec, Apache
members who are also OSGi members will have access to people, processes,
and information than non-OSGi Alliance members will not, giving them an
advantage to implementing and shaping future evolution. There is no way to
level the playing field between the two. Which would be true of any
standards body where you have to be a member to participate.
Shaping future evolution is definitely a privilege of OSGi Alliance members
and again, I have no problem with that part. The problem is not whether
OSGi members have more information or not, it's whether non OSGi members
have enough information or not to implement a spec. When the
implementation is used as a play ground for the spec design, that's
definitely not true.
Whether future evolution is long-term (i.e. released specs) or
short-term (i.e., draft specs), Apache members have no real say in
shaping them, so I don't see the issue.
You seem to be overly concerned about a case where either a draft spec
is not available or is somehow out of date. Admittedly, someone without
access cannot easily hit the target, but it doesn't really matter since
they can get a new target as soon as the draft is updated. He or she
still has no input on what the updated draft will be. Oh well.
And even is some OSGi Alliance member and Apache committer does have
access to unpublished drafts, who cares? It still benefits us a
community to see the working being done and be exposed to it before the
draft is updated, if for no other reasons than the ones stated above.
-> richard
