On Mar 18, 2008, at 10:50 AM, Mike Heath wrote:
Alan D. Cabrera wrote:
On Mar 5, 2008, at 9:03 PM, Mike Heath wrote:
Alan D. Cabrera wrote:
<snip>
This seems like a good idea. I have some questions.
When we cut a release of this code, what version will it be? What
will
be its Maven group and artifact id?
What about the other AsyncWeb client? It looks like people are
modifying that quite heavily. Are we going to need to do a pre-2.0
release of that as well?
Now you're asking hard questions that I'm not sure I have a good
answer
for. I think this will take some discussing.
To get the discussion started, I'll suggest that for AHC we use the
Maven group 'org.apache.asyncweb' and for the artifact id we use
'ahc'.
For the version, how about 1.0?
For AsyncWeb client, I think we should use the group
'org.apache.asyncweb' and the artifact id 'client'.
Seems good to me.
What about the work that's currently being done on the "old" asyncweb
client? What are the plans for that? I ask about this because it
looks
like someone is actively working on it. Will we also have a 1.0
release
of org.apache.asyncweb:client?
I think we should move the "old" asyncweb client (a.k.a. AHC) over
to a
branch in AsyncWeb and continue to maintain it there.
Was this ever released?
I think we should release all of AsyncWeb (client, server, codec,
extras) together as a 1.0 release. Because both client and server
depend heavily on AsyncWeb commons, this makes sense, IMO.
When you speak of client, do you mean the "old" one or the "Geronimo"
one?
In the AsyncWeb client project, I would like to move to the API that
we
proposed earlier and was discussed on the mailing list. Having code
that everyone can see and tinker with will make it easier to
facilitate
discussion. It's going to take a lot of work and creativity to come
up
with an API that can accomplish all the things we've been discussing
as
well as remain consistent between the client and server sides of
AsyncWeb.
Good idea. Please see an earlier thread that was started about how we
can proceed on this.
So to summarize:
- We move AHC from Geronimo sandbox to a branch in AsyncWeb and
maintain it from there (I would like to see an AHC release soon too.)
- For AHC we use the group name o.a.asyncweb, the artifact id 'ahc'
and
the version 1.0
- For the new AsyncWeb client we use the group name o.a.asyncweb and
the artifact id 'client' this will also have the version number '1.0'
and will be released with the collective AsyncWeb project.
I think that we should make it version 2.0. It fits nicely with its
MINA 2.0 ties and it more clearly delineates it from the 1.0 release
that we're proposing. IMO, simply renaming the artifact id, while
still a good idea, is not enough to differentiate the new API.
- I'll move the new client API we've been discussing into AsyncWeb
client so we start developing it and continue discussing it
Please just move the interfaces, per the other discussion on how to
proceed, so the community can start submitting examples based on the
use cases.
It would be good if the HttpConnector was an HttpConnectionFactory w/
out the HttpClient factory methods as we discussed. If you still do
not agree then I think it's best that we wait until we reach a
consensus on these far reaching changes. I believe that I am still
waiting for your reply on a number of issues no that old thread.
Is everyone ok if we move forward with this plan? Do we need to call
for a vote?
Wait 72 hours for the discussion to sink in and then call a vote.
Regards,
Alan
