Hi all,
After an
heavy WE of works
and discussion with Sébastien, here are an outline of the commits (BTW,
everythink I did is traced in the Papyrus bugzilla). The intend of the
work was
to try to reconcile my contribution to the Papyrus backbone and the
contributions of ATOS in a way everyone could be satisfy. The refactor
done in
v656 put me in trouble because there was too many changes (mainly code
reformating) and I was completely lost in the new code.
The first
important result
is that a priori everything is working again now. The second important
result is
also that now I have a clearest view of the behaviorals changes made
around the ATOS
refactor for
supporting the control mode. More comments in lines below.
BTW, in
accord with Sébastien, we propose to have another telco Wednesday
morning (10 am to 11:30 am) in order to
discuss and plan the incoming actions required for the backbone.
Finally, I
think that this fight should stop and we have to work in a more
collaborative and productive manner.
Cedric
Raphael FAUDOU wrote:
Yes, it
compiles but we
lost many things :
- all the work about dead code (unused tags) => 1 week
effort from thomas lost
The unused tags are
not yet back
because they are too many, and mainly because there were written as a
java
comment just before the methods rather than be added directly in
javadoc. I
commit to clean
the code for the end of the year. I will do that certainly within the
Christmas
vacation. It will be completed at last for the first week of January
2010.
- dirty mode fixed by Thomas does not work anymore => 4
days effort lost and not usable
About the dirty
mode, there are
two remaining issues: (i) composite diagram needs to be modified in
order to
follow the new way to save model as introduced in the refactor done in
v656.;
(ii) it seems also there is some side effect on the activity diagram
that
cannot be closed.
In addition, I
think that this refactoring
point has indeed introduced a too tightly coupling between the core and
the nested
editors. These latter should now implement some methods in a particular
way
(doSave() and isDirty()). The core should support any kind of editors,
even
those that do not implements these methods or implements them in their
Eclipse
way. So this change is not compatible with the spirit of the core (not
dedicated to GMF or Papyrus). I propose to work with Thomas to find a
more flexible
solution and to discuss this point next Wednesday.
- Thomas name has disappeard from headers => completely
innaceptable
This comes from the
way the
restore have been done. All citations (2 instances) are restored. My
apologies
for that.
We have the
feeling to
have come one month back and it is a very disappointing situation :-(
We are blocked for our work on the SysML profile support and for BDD
and IBD
diagrams.
This is a red signal as we can not
continue
working like that (frustration, much time consumed).
As immediate actions, I request that :
1/ the backbone code be reverted from two days ago.
Now that I have
tracked and
understand the modifications done around the v656 refactor, I have
restored
these modifications. But, I think we should reconsider the way they
have been
done, and propose more universal solutions.
2/ control
mode fixed as
it is a blocking point for collaborative work
The control mode is
restored,
but we should consider to change the way it is coupled to the backbone.
I
propose to put the controlmode as a service of the backbone, and let
the sashdi provide required functionalities, but on a higher level than
the
one actually provided. => bug
3/ backbone
evolutions
done by Cedric be commited on a branch and not on the head
On this point, we
have concluded
with Sébastien that all the "unused code" will be put on a separate
branch
in my local workspace. It will be done as already said previously at
last within
the Christmas vacation. However, concerning future works, we (Sébastien
and me)
think that there are two possibilities that indeed should apply to all
papyrusdevs:
1) either a new functionality get a consensus (every Papyrus partners
agree on the
new functionally, in this case the related code has to be of course
part of the
Papyrus core plug-ins; 2) the proposed functionality is not validated
by all. In
this latter case, the functionality has to be introduced as an option
of the tool
which is disable by default .
Thanks
raphaËl
Cedric Dumoulin a écrit :
Hi,
The head compils and runs again, except the controlMode plugin and the
dirtyFlag behavior.
I will check that asap (tomorrow) and try to find a solution.
Cedric
Cedric Dumoulin wrote:
Hi all,
Please do not commit on the core, especially on backbone, has I am
currently trying to recover it .
Cedric
Thibault LANDRE wrote:
Hi all,
A commit made yesterday (r734) breaks the compilation of the project.
The class NavigatorUtils needs some methods from the class EditorUtils
that
have disappeared.
It seems that the commit made yesterday was made from a code that
wasn't
updated.
If you want to work with code from a specific revision, please create a
branch.
Another point, before comiting, please ensure that you are up to date
with the
code on the repository.
Finally, I think it is safier that every developer, specifically on the
backbone, get all the source code of the project in their workspace to
check if
their modifications don't impact other parts of the project.
Regards,
Thibault
_______________________________________________
mdt-papyrus.dev mailing list
mdt-papyrus.dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mdt-papyrus.dev
_______________________________________________
mdt-papyrus.dev mailing list
mdt-papyrus.dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mdt-papyrus.dev
_______________________________________________
mdt-papyrus.dev mailing list
mdt-papyrus.dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mdt-papyrus.dev