Comments inline.
Luciano Resende wrote:
So, just to make clear, the plan here, while using sca-contribution,
is to
call SCADomain.newInstance(String DomainURI) and then we would require
sca-contribution to be available, so we can find the root of the
contribution, and then we would get deployables from that ?
Yes
What should be the behaviour if no sca-contribution is specified ? the
SCA
spec does mention this file is optional. Also, what we do if no
deployables
is specified ?
If you don't give any deployable composites to newInstance and you don't
have any deployables in an sca-contribution.xml, you end up with an
empty domain, that's all.
For a sample using sca-contribution.xml, what do you want to
demonstrate in
particular with this sample ?
How about specifying deployables :) I see two interesting use cases:
- the webapp sample
- the implementation-composite sample, where you'll list only the outer
composite as deployable, showing that the inner composite is not a
deployable.
On 5/15/07, ant elder <[EMAIL PROTECTED]> wrote:
On 5/15/07, Jean-Sebastien Delfino <[EMAIL PROTECTED]> wrote:
>
> [snip]
> ant elder wrote:
> > Here's todays weekly IRC chat log. The only topic discussed was the
> > SCA 0.90release, the current plan is to cut a
> > 0.90 release branch this Wednesday.
> >
> > We went through the open JIRA's targeted at 0.90 to see if anyone
wanted
> > particular JIRA's as must fix for 0.90 and which ones could be
> > defered. The
> > current JIRA's for 0.90 are:
> >
>
http://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&mode=hide&sorter/order=DESC&sorter/field=priority&resolution=-1&pid=12310210&fixfor=12312478
> >
> > . Tomorrow I'm going to move all of those out of 0.90 except for
> > TUSCANY-1247, TUSCANY-1248, and TUSCANY-1265. If anyone has any
others
> > which should be in 0.90 just say.
> >
> > The samples were also discussed, sounds like most of them are ready,
> > there's
> > some issues around the those related to how to write Tuscany
extensions,
> > discussion will continue on the mailing list.
> >
> > Venkat will run RAT against the current distro's and email the
mailing
> > list
> > the results, slaws will try the distro's on linux.
> >
> > The latest distributions built of the trunk code as of just now
> (r537917)
> > are at:
> > http://people.apache.org/~antelder/tuscany/latest/<
> http://people.apache.org/%7Eantelder/tuscany/latest/>.
> >
> > Please try these out and report any issues (or even just that you
> > tried and
> > a sample worked!)
> >
> > ...ant
> >
>
> Sorry that I missed yesterday's IRC chat. I have a few more things to
> add to our TODO list for the release:
>
> I'd like to have in the binary distro a JAR containing the Tuscany
> source that people can attach to the Tuscany JAR for debugging in
an IDE
> (we already discussed this on the list but can't find the thread now).
Could just the src distribution be used for this for 0.90? I've just
tried
and Eclipse works fine using that as the src for the Tuscany all jar.
> I am not sure where we are with our Webapp story. If we go with the
> context listener approach, we need to adjust it to take its
> configuration and list of deployables from sca-contribution.xml
instead
> of web.xml. If we go with another approach we need to adjust the
sample.
>
> For consistency, SCADomain.newInstance() should also work without a
list
> of composites and take its configuration from sca-contribution.xml if
> there is one.
>
> We need some samples to show sca-contribution.xml (adjust some of the
> existing samples).
>
> The above items are important IMO as they touch APIs used by
application
> developers.
I've taken all the config based on init-params out of the webapp host so
currently it just uses whatever composites it finds in the top-level
classes
folder of the webapp. The calculator-web sample works with this and
doesn't
specify any special config. is this not enough for 0.90?
Making all composites deployable is going to break in most cases. If you
have 2 composites A and B for example:
- if A includes B, then you'll get an error (name collision) if you try
to add both A and B to the domain
- if A uses B as an implementation, then you'll get unexpected
application behavior as the intent of the application developer is to
use B as a nested implementation and not a set of top level components.
Also, if you intented to use a particular contribution as a vehicle for
implementation artifacts (composites + other things) then chances are
that you won't have an sca-contribution.xml file in it, and
automatically adding all composites to the domain and activating all the
components they contain at the domain level is going to be pretty
unexpected as well.
If not then is
there anyone who volunteers to add the sca-contribution.xml (and when
can
they get it done?) ?
...ant
--
Jean-Sebastien
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
