Dear Ed W.
Thanks for your note. The slides I sent you where about another
topic, and do not cover the topic we will contribute to MXF. We will
show that in March in the OMG/Eclipse symposium.
As well, I'd like to clarify, that MXF stands for Model Execution
Framework, not Model Transformation ;-) We do not intend to
contribute anything to model transformation, we use QVTO for this at
the moment.
I liked the MXF proposal details, and we will remain in synch with
the original proposal, and as well welcome, the original
implementation as one possible way to do a model execution
framework. We have a lot of respect from the work of the original
contributors, and I offered them our help at the beginning -
unfortunately never got a reply.
Unlike the original implementation, we will propose a leaner
implementation, that uses OCL to define actions, and that will
reduce actions to "one big step" without intermediate states: this
is XOCL.
Regards, Philipp
On 10.01.2012 09:12, Ed Willink wrote:
Hi
I'm looking forward to a presentation of XOCL, as part of the
joint Eclipse/OMG symposium just before EclipseCon, as a chance to
try to understand what XOCL is really about. I've seen the
"Montages AG - Business Modelling Practice and Innovations" slides
on which there is a nice two dimensional editor for model
instances, but beyond that it was not clear to me what was new in
comparison to what has happened in parallel. Discussions on OCL
Analysis have provided some strong motivation for promoting the
OCL Impact Analyzer from examples in the Juno release, so that an
independent development can be replaced by a 'standard' one.
'Model Transformation Framework' is a wonderful term that can mean
whatever you want. When the original MXF was proposed I was
enthused, until I read the proposal detail and found that it was
nothing like what I was hoping for. I am sure that the project
name can be reused for a variety of purposes, but I think it is
unfair to burden any new project with the first couple of years of
misleading history.
In the meantime, as Ed Merks mentioned, Xcore/Xbase provides an
extended Ecore framework. With the advent of direct OCL 2 Java
code generation for the dispatch table based OCL Virtual Machine,
the OCL VM forms the root of another model transformation
framework that can be extended to support QVT and other
approaches.
"From our side we will contribute one MXF framework called XOCL,
which is simply a set of standardized OCL annotations for ECore
models. This is, as Ed Merks mentions simply a usage of existing
stuff, not much new. "
This suggests that the new project is more like a library than a
tool. However the slides introduce both MCore and XOCL.
Unfortunately, as with many PPTs, it is difficult to grasp quite
what is going on without the presenter's words and pace. MCore
appears to be much more than a library; is it part of the
contribution?
The slides conclude with "MCore maps back to ECore + OCL (XOCL)
and can be considered as a simplification of modeling with ECore
and OCL". I would like to understand how this compares with the
OCLinEcore editor and its underlying use of Delegates that were
probably not available when the XOCL work was started.
I welcome anything that adds to the capabilities of modeling and
OCL in particular, especially anything that adds manpower, however
I think the alignment with current projects needs to be clarified
and a clear scope and name for a new project identified. Perhaps a
meeting at EclipseCon may be helpful.
Regards
Ed Willink
On 10/01/2012 07:17, Ed Merks wrote:
Philipp,
I don't imagine what you're planning fits exactly the scope
that's been spelled out. Certainly things have evolved, as you
know, since that scope for MXF was written, i.e., the
introduction of delegates for operations, constraints, and
derived features in EMF. The combination of these things allow
behavioral aspects to be defined directly in the Ecore model in
an extensible way that supports languages like OCL. I'd rather
see things like XOCL be part of the OCL project than to revivew
a stillborn cross cutting project. Better the OCL project
diversify...
The new Xcore work is also about model execution (for Ecore), to
some extent, but I'd rather keep that as part of the EMF
project, not move it to a cross cutting project.
I'm not sure how the other PMC members feel about this. In
general we have a large number of dead project that need
cleaning up. Personally, in the future, I'd rather see more
life injected into projects that are currently alive.
Regards,
Ed
On 10/01/2012 6:50 AM, Philipp W. Kutter | Montages AG wrote:
Dear Wayne.
Thanks for the clear directions, we will follow them.
I will start by discussing with prospective Architecture
Council mentors, for the topic at hand and then follow their
advice.
Where is the list of the Architecture Council members, and
which projects they already mentor?
Regards, Philipp
On 09.01.2012 20:15, Wayne Beaton wrote:
It seems that by not speaking, the project has spoken. Or
something to that effect.
Now it's in the Modeling PMC's hands. With their unanimous
consent, we can change the project lead and committers. The
easiest thing to do is to replace the project lead and have
the new lead retire the existing committers and nominate the
replacement committers via the portal.
The Modeling PMC has to have a transparent discussion about
this. This discussion--which can be initiated by anyone
(either a member of the PMC, or somebody like
Philipp)--should include a few words stating that the
project team has become unresponsive and that another party
has stepped forward to take the helm. The discussion should
include some indication of confidence that the new project
team is ready for the responsibility in terms of
understanding the EDP, working in open source, etc. followed
by a minimum of three +1s and no -1s from the PMC.
My records show that the project is in incubation, but has
no mentors assigned. As part of this reassignment, I'd like
to see at least one Architecture Council mentor identified
for the project.
Thanks,
Wayne
On 01/05/2012 11:44 AM, Philipp W. Kutter | Montages AG
wrote:
Agree 100%
As you wrote on 26.4. that you will check with them the
status, I assumed, that the fact that the project is
unresponsive is already here.
How long do we want to wait?
mxf-dev@xxxxxxxxxxx
should reach the original people, no? Hello: anyone out
there???
Regards, Philipp
On 05.01.2012 15:58, Wayne Beaton wrote:
How does the existing project team feel
about this?
The easiest way to proceed is for the existing project
team to accept your XOCL contribution, move it into the
IP process, and initiate committer elections for the new
developers (citing the contribution as the required
demonstration of merit). Once on board, you can nominate
and elect a new project lead. That lead can retire the
inactive committers. The existing project lead can
retire by sending me a note.
That's the ideal.
If the project team is unresponsive, the Modeling PMC
can--after transparent discussion and unanimous
consent--decide to replace the project lead and
committers.
Make sense?
Wayne
On 01/05/2012 09:03 AM, Philipp W. Kutter | Montages AG
wrote:
Dear Wayne.
I have not seen anything since April now. I assume
thus that the project will be either closed, or should
be taken over from another party.
In the meantime we increased activities on our own
model execution framework, and we definitively would
like to take over the project. I cc'd their mail list
to see any reaction from the original people.
The scope of the project needs not be changed, as they
positioned it as an open project, allowing to welcome
all MXF, not only the original proposed one. Thus we
will be open for the original contributions, and
others coming from the TopcaseD area (see discussion
on mail list).
In additon to the original scope, we will much more be
focused on project collaboration with other Modeling
projects, mainly those implementing OMG standards,
such as ECore, OCL, QVTO, Acceleo, and DI from
TopcaseD. Here the points we will bring to the scene:
- ECore will be the basis for all metamodels, such
that other modeling projects for persistence (such as
CDO) and different ways to express syntax (visual,
textual, tree/table) can be added easily
- Reuse of _expression_ languages of other projects
(OCL, imperative extension of OCL from QVTO, and newer
ones like XBase)
We especially intend to use the project to make sure
that topics such as dynamic/static binding of
operation calls, overriding/overloading, multiple
inheritance are solved the same way as in ECore/Java.
(we filed Bugzillas for this topic in the OCL project,
which where already partially fixed)
From our side we will contribute one MXF framework
called XOCL, which is simply a set of standardized OCL
annotations for ECore models. This is, as Ed Merks
mentions simply a usage of existing stuff, not much
new.
Michael Golubev will bring the knowledge to the scene,
how to do the builds and will help me to follow all
the Eclipse processes. He is the component lead for
GMF Tooling and UML2 Tools.
Thus: there needs nothing to be added to the original
plan.
Please let us know how to proceed.
Regards,
Philipp
On 26.04.2011 14:54, Wayne Beaton wrote:
Hi Philipp./
The project appears to be dead on arrival :-)
I will check with the PMC and project founders to
see what their plans
are. Hopefully you'll see some activity from the
project.
Wayne
On 04/26/2011 05:16 AM, Philipp W. Kutter wrote:
Dear Anne.
Has there been any news since 7.4.2009?
I have neither seen the Eclipse page, nor the
initial code contribution.
Any input welcome. I will as well try to contact
the founders of the
project as soon as I find time.
Regards,
Philipp
Am 07.04.2009 18:47, schrieb Anne Jacko:
Hello all,
Since there has *not* been a request from a
member of the Eclipse
community to hold this review on a conference
call, there will be no
Review Call tomorrow (April 8, 2009).
The EMO has declared this review to be
successful based on the review
docuware and on community feedback.
Congratulations to the MXF team on
their successful review.
Please contact emo@xxxxxxxxxxx
with any questions. Thanks.
|
