Christophe,
Thank you for providing feedback to our report. Please find my
comments below:
For 2011, the effort estimation is 24 PY. Does this mean
that the participating companies would have to invest
24/<NbOfInvolvedCompanies> full time person during one
year,
No, this wouldn't be feasible. Even if we had enough funding it
would not be realistic to assume that we could put a team of 24
people across different companies together and make sure that they
collaborate in a coordinated and efficient way. There needs to be a
significantly smaller group and some sort of ramp up to get this
successfully started.
or are you considering that a part of the work will be
managed by the current development
teams?
It is sure that the current development teams will continue to do
things in 2011 even if no funding at all would come from the
Modeling Platform Working Group. But there wouldn't be any guarantee
as to which extend these works would cover the Modeling Platform
requirements and when the results would be shipped.
In the second case, could it be possible to know (more or
less, I know it is Open Source), the extra effort the
participating companies would have to provide?
The gap analysis, which is the first part of the planning study,
indicates only the efforts that are required to be made regardless
of who will finally take charge them and who will provide the
necessary funding. In the second part of the planning study we will
work on a roadmap in which we are going to address these aspects,
i.e., which implementer companies would participate, which is the
realistic number of resources that each of them could contribute,
which of the prio 1 work packages could be realized, and which
deliverables and milestones could be scheduled.
But again, for the sake of providing a project plan that is
realistic and executable, we cannot base it on results which may or
may not be achieved in the open source space but is not really under
our control.
Showing these figures to my management will not have the
expected result. Saying that we will have to provide/pay for 6
persons for this project will never be accepted.
We are aware of this problem and therefore will think of a way of
providing a useful result without requiring an investment as high as
what the gap analysis currently indicates. The roadmap which we will
ship by end of October will propose such a way.
Regarding the estimation itself, am I wrong saying that
your estimation and proposal is strongly coupled with a
repository provider?
What do you mean by repository provider? Is it a model repository
like CDO?
In the very first versions of the requirements, it was
wished to be “Loosely
coupled
to different repository providers” (line 49 of the
Excel sheet of the 24/02/2010 sent by Ian).
Assuming that the answer to my question above it yes, I'd say that
this requirement correspond to requirement 1.3 + E-Config-1 "Support
for different model repository types" in A1 "Version Management" in
the gap analysis spraedsheets. The detailed capabilities of this
requirement explain what kind of repository types are to be
addressed.
Is it planned in the current architecture to be able to
easily change some part of the platform?
Clearly yes. This includes the capacity of swichting between
different model repository types as well as supporting model
repositories, files-based SCMs or both in parallel. However, from
other feedback which we have got so far we have concluded that this
is not going to be a no 1 priority. The focus of the first iteration
therefore would be on supporting one model repository type, namely
CDO, and to some extend also file-based SCM systems.
Thanks and regards,
Christophe
PS: I read Ian’s mail about prioritization, but this will
just stretch the project over time.
What we can learn from the gap analysis already now is that the gap
between exisiting Eclipse modeling technology and the Modeling
Platform requirements is pretty big (and maybe significantly bigger
than many of us had hoped). It therefore will not be realistic to
assume that ALL capabilities that are currently have a prio 1 can be
realized within 2011. It will be unavoidable to make choices, i.e.,
reprioritize the prio 1 capabilities, and yes, decide which of them
can be left to a later iteration.
I all the same believe that it will be possible to identify a subset
of the prio 1 capabilities that fits into the 2011 schedule and will
make up something useful. We then would be able to work on that
subset in a coordinated way and address success-critical aspects
like consistent integration to a complete platform, system testing
based on real world use cases, etc. in an appropriate manners. So,
compared to trying to do more that what is achievable and ending up
with nothing, we will eventually have more by doing less.
Kind regards,
Stephan
_______________________________________________
mpwg mailing list
mpwg@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mpwg
--
Dr. Stephan EBERLE
----------------------------------------------------------------
Product Development Manager
6 avenue de Verdun
92250 La Garenne Colombes, France
Phone : +33 (0)1.73.54.00.30
Fax : +33 (0)1.73.54.00.32
Mobile : +33 (0)6.64.18.39.10
E-mail : stephan.eberle@xxxxxxxxxxx
www : http://www.geensys.com
----------------------------------------------------------------
|