Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [Fwd: Re: [jwt-dev] refactoring]

Hi Chris, Florian,

That's indeed good information & work.

Alas the build logs I'm getting are really depressing right now, do you feel like you're making any inroads there ?

Regards,
Marc

Christian Saad a écrit :
Hi Florian,

yeah, we made some progress in fixing the build but it's really something which (I can only guess) often results in bite marks on the table ;) I hope the FactoryRegistry is fine now, I implemented it because I didn't find a patch although I guess it would be a good idea to go through the bugs and integrate the patches which are present there.

I put a description of the extension point to the wiki, which is now again a bit extended: I took the opportunity to further refactor it removing the external view descriptor (name & icon can now be given directly in the extension point) and added an entry for the ID (which is used instead of the id in the view file). I think this will make it easier for others to define views and for ourselves if we at one time change the *.view file to a simpler format. Also, there is no distinction between internal & external views anymore (internal views are now registered in WE's plugin.xml) which will make it easier to remove the default views.

Thanks also for the clarification on the EPC view elements!

Best Regards,
Chris

Am 08.05.2010 19:09, schrieb Florian Lautenbacher:
Hi Marc, hi Chris,

wow, kind a long discussion. Sorry for not being able to answer earlier and to help to fix the build problems. It seems there was already some progress in fixing the build problems, but not everything is fixed, since Hudson is
down again.
Great to hear that the FactoryRegistry has been fixed. I noticed the problem that only one view is completely loaded, but not the other already some time ago and I thought I created a patch for it, but I can also be wrong - maybe
I didn't upload it. But now its finally fixed! Thanks for that, Chris!

Chris, could you point to a wiki page where users can find a description how the factory registry is bound to a new view? Is it by specifying it in the
plugin.xml as we had it already or do you use a new mechanism?

Concerning the problem with opening a file with an EPC view: I can confirm that we once already had such a problem when changing a file and opening it
after restarting again. But I never tried to load it from the project
workspace, so I guess this is an "old" bug.

Concerning the figures of the EPC view:
- the initial node and final node are displayed in the EPC view as a kind of
event as there is no distinction for EPCs between such events and others
which is why I didn't create an entry in the palette for these elements.
- Roles and Data are dynamic entries in the palette and there is already an open bug for this extension as far as I remember. I only added the static elements in the palette in the first place, but you're right, those should
be added, too.

I hope that information helps a bit.

Best regards,

Florian


-----Ursprüngliche Nachricht-----
Von: jwt-dev-bounces@xxxxxxxxxxx [mailto:jwt-dev-bounces@xxxxxxxxxxx] Im
Auftrag von Christian Saad
Gesendet: Freitag, 7. Mai 2010 16:54
An: Java Workflow Tooling
Betreff: Re: [Fwd: Re: [jwt-dev] refactoring]

Hm, weren't there a lot of dependencies? ;)

Another thing:

Since the build server is down (again) and I noticed that the
FactoryRegistry isn't working as intended (it seems to just load the
first registry it finds at the extension point. I noticed this when the
EPC view did not work since it apparently quit loading when it found the
UML view factoryregistry) I did a bit of rewriting (which we wanted to
do anyway at some time). FactoryRegistries are now bound to a certain
view in the extension point (if not set the default registry is used).
This also provides (reliable) automatic switching when the view is
changed and one doesn't have to make case distinctions when implementing
custom registries.

But before I upload the code I have a question: When enriching a
workflow with the EPC aspect, then quitting and restarting the workbench
I get an error that the epc conf wasn't found and the workflow cannot be
loaded. If I then open it from the project workspace, everything is ok.
Is this a new problem which I introduced or can you confirm this for the
old version too?

Another question (perhaps for Florian): The EPC view defines figures for
roles, data, initial/finalnode,... but somehow it seems they aren't
added to the palette.

Regards,
Chris

Am 07.05.2010 14:14, schrieb Marc Dutoo:
You could also try to temporarily drop all conf packages and samples,
just to see if it works. The dichotomy method :)

Regards,
Marc

Christian Saad a écrit :
Hi Marc,

I don't think it's usually a problem but the
org.eclipse.jwt.we.conf.aspects package seems to be a possible
candidate where things go wrong, especially since most of the compiler
errors reported refer to the conf model/edit plugins. Of course this
could be just a coincidence but I think currently the only thing left
to try out and if I understood Mickael right its also in line with
what he suggested.

If it doesn't work, I'll try to ask on the mailing list.

Regards,
Chris

Am 07.05.2010 12:46, schrieb Marc Dutoo:
Hi Chris

I think that if having a org.eclipse.jwt.we.conf on one side and several org.eclipse.jwt.we.conf.* (ex. ui) on the other side is a problem, there
it is far from being the only one of this type, so I'm a little bit
doubtful.

But you're the guy on the deck, so do whatever you feel can help you
solve this, and we'll make it better afterwards.
And looking at Mickael's opinion, it seems we're still not enforcing his
"all classes of a x.y bundle should go in an x.y package or
subpackages".

Is there a mailing list for athena / cbi ?

Regards,
Marc

Christian Saad a écrit :
I removed all the reexports because I tried to eliminate possible
causes of this error. Generally I'm a bit cautious using those because
it hides dependencies from the developer but anyway obviously this
hasn't been the problem here.

Would it be ok for you if I added a unique package to conf model&edit,
e.g. model stuff goes to org.eclipse.jwt.we.conf.model.* and edit
stuff to org.eclipse.jwt.we.conf.edit.*

And as said, if this doesn't work... the only thing remaining is to
make an offering to the Eclipse god...

Regards,
Chris

Am 07.05.2010 11:50, schrieb Marc Dutoo:
Hi Chris

I see you're making progress.

Why not add more visibility:=reexport in manifests ? It's an easy
way to
solve a lot of problems, and quite logical in a lot of case, for
instance you can't expect people using aspects without using EMF.

Your opinion ?

Regards,
Marc

Christian Saad a écrit :
Hi Marc,

thanks for fixing this so quickly, alas, Hudson is down again atm....

I've checked out your changes and also for me they don't give any
errors. I've refactored the helpers plugins but haven't uploaded them
yet as I'd like to do some local testing first.

Regards,
Chris

PS: jdom is used by the WE (to parse view files)

Am 06.05.2010 17:29, schrieb Marc Dutoo:
Hi

I've downloaded the Eclipse 3.5 Helios r2 updates, checked out
org.apache.xalan from Orbit. Now, no more build errors, and I've
successfully tested my committed refactorings.

I've also update the http://wiki.eclipse.org/JWT_CVS pge describing
how
to set up a development environment :
* added dependencies : ATL plugin, org.apache.xalan from Orbit
* TODO is jdom still used, where ?

Regards,
Marc

Marc Dutoo a écrit :
Hi

Opened a bug about it :
https://bugs.eclipse.org/bugs/show_bug.cgi?id=311878

My analysis of the package conflicts :

*

jwt-we-conf-model.edit/org.eclipse.jwt.we.conf.aspects.event.provider

: already in jwt-we-conf-model, moved to .provider, no impact
because
was obsolete and not exported DONE
*
jwt-we-conf-property-model/org.eclipse.jwt.we.conf.aspects.factory
:
already in jwt-we-conf-model, moved to .property, 1 impact in
jwt-we-conf-property-model.edit (as well as in manifest and
plugin.xml) DONE
* jwt-we-sample-staticaspect/org.jwt.sample.logging : just been
removed, in duplicate with those in jwt-we-sample-logging DONE
* jwt-we-helpers-application/org.eclipse.jwt.we : already in jwt-we
STILL TODO



Still TODO : separate
jwt-we-helpers-application/org.eclipse.jwt.we,
test build and app


I'm sorry, but right now I've got a lot of build errors in my
workspace (ex in org.eclipse.jwt.converter.Converter : can't see
import org.eclipse.emf.common.util.URI ), possibly because I'm not
using Eclipse 3.5 r2 (so no latest atl plugins etc.).

Though I've already committed the items marked DONE above, since
they
have almost no impact, dear Chris, could you please test my changes
for me, in addition to renaming
jwt-we-helpers-application/org.eclipse.jwt.we ?

(obviously, I'll try having a workspace in order at some point -_-)

Regards,
Marc

Marc Dutoo a écrit :
Hi Chris, Mickael

Mickael, thanks for your expert inputs.

Since it is a bad practice, I guess I should do the refactoring
sooner rather than later, though I'd prefer have a bigger error
margin. I'll try having it done and tested before 16:00. I'll keep
you in touch.

Regards,
Marc

Christian Saad a écrit :
Hi Mickael,

great to hear from you :)

Thanks for the confirmation! After removing the reexports, I had
one
good run (more or less, if it gets through then it currently
fails
at the tests) and now again having problems (see below), e.g.
with
the

org.eclipse.jwt.we.conf.edit/src/org/eclipse/jwt/we/conf/aspects/event



package which is definitely shared between two plugins (and also
some others).

Marc, would it be possible to resolve this more or less
easily? (I
would prefer putting the files in different packages rather than trying to get it to work the way it currently is because I've no
idea which part of the build causes this problem and if it
responds
to manifest declarations or not)

Regards,
Chris


@dot:
[mkdir] Created dir:

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/
I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/@dot




[javac] Compiling 24 source files to

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/
I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/@dot




[javac] ----------
[javac] 1. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/event/AspectEventManagerNotifyChangedListener.java



(at line 35)
[javac] private final AspectEventManager aspectEventAdapter = new
AspectEventManager();
[javac] ^^^^^^^^^^^^^^^^^^
[javac] AspectEventManager cannot be resolved to a type
[javac] ----------
[javac] 2. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/event/AspectEventManagerNotifyChangedListener.java



(at line 35)
[javac] private final AspectEventManager aspectEventAdapter = new
AspectEventManager();
[javac] ^^^^^^^^^^^^^^^^^^
[javac] AspectEventManager cannot be resolved to a type
[javac] ----------
[javac] 3. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/event/AspectEventManagerNotifyChangedListener.java



(at line 38)
[javac] this.aspectEventAdapter.notifyChanged(notification);
[javac] ^^^^^^^^^^^^^^^^^^^^^^^
[javac] AspectEventManager cannot be resolved to a type
[javac] ----------
[javac] ----------
[javac] 4. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/provider/AspectEnrichedItemProviderAdapter.java



(at line 34)
[javac] import org.eclipse.jwt.we.conf.Aspect;
[javac] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[javac] The import org.eclipse.jwt.we.conf.Aspect cannot be
resolved
[javac] ----------
[javac] 5. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/provider/AspectEnrichedItemProviderAdapter.java



(at line 35)
[javac] import org.eclipse.jwt.we.conf.AspectInstance;
[javac] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[javac] The import org.eclipse.jwt.we.conf.AspectInstance
cannot be
resolved
[javac] ----------
[javac] 6. ERROR in

/opt/users/hudsonbuild/.hudson/jobs/cbi-soa-jwt-integration/workspace/build/ I201005060556/eclipse/plugins/org.eclipse.jwt.we.conf.edit/src/org/eclipse/j
wt/we/conf/aspects/provider/AspectEnrichedItemProviderAdapter.java



(at line 36)
[javac] import org.eclipse.jwt.we.conf.ConfPackage;
[javac] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[javac] The import org.eclipse.jwt.we.conf.ConfPackage cannot be
resolved



Am 06.05.2010 11:31, schrieb Mickael Istria:
Hi,

You're right, it's weird anyway because it did work before, at
least
Mickael built it this way I guess, maybe he's got an idea ?
This is probably because CBI build is p2-based (uses
p2.gathering=true
property), and is more rigorous, and far better, about
dependency
resolution, whereas the old builder was not p2-based (standard old-school build + p2 generation on the old-school update-site).

So maybe don't change too much things yet, just enough to now
whether
that's the right solution or not...
I think having the same package in several bundles is indeed the
cause
of the problem. That's one of OSGi worst practices, so that it
can be
the cause of such troubles. Basically for a bundle
org.eclipse.jwt.my_bundle, all classes inside of it must be in a
org.eclipse.jwt.my_bundle or child package. Otherwise, it
could be
confusing for developers, install-time resolver (p2) and runtine
resolver (Equinox).

Regards,

--

Mickael Istria
R&D Engineer
BonitaSoft - Open your processes<http://www.bonitasoft.com>
email
: mickael.istria@xxxxxxxxxxxxxx


This message and any attachment (the "message") is intended
solely for
the addressees and is confidential. If you receive this
message by
mistake, please delete it and notify the sender immediately. Any
use not
in accordance with its purpose, any out-spread or disclosure,
either as
a whole or partially, is prohibited except with formal approval.
Internet cannot guarantee the integrity of this message,
therefore
BonitaSoft will not be liable for the message if modified.



_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev

_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev

_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev


_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev
_______________________________________________
jwt-dev mailing list
jwt-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jwt-dev


Back to the top