Hi
You should see
8e6d is your original contribution.
8455 was my initial fixes.
96ed was your further work.
b1f5 is my enhancement to keep the ClassDatum analyses on a
TypedModel basis so that endogeneous transformations work. This is
accompanied by the addition of a ClassDatum.typedModel property.
But for the most what you see is the QVTp2QVTg that you contributed;
thanks, very helpful. The refactored QVTs does not exist. The
classes in org.eclipse.qvtd.compiler.internal.scheduler await a
refactor to make their information visible as an Ecore model.
Potentially
org.eclipse.qvtd.compiler.internal.schedule2qvti renames as
org.eclipse.qvtd.compiler.internal.qvts2qvti
and
org.eclipse.qvtd.compiler.internal.scheduler renames as
org.eclipse.qvtd.compiler.internal.qvtp2qvts
with the 'model' from org.eclipse.qvtd.compiler.internal.scheduler
extracted and merged with org.eclipse.qvtd.pivot.schedule.utilities
as
org.eclipse.qvtd.pivot.qvtschedule
I took a quick shot at this on Tuesday but the 'model' classes are
currently too rich to Ecore-ify trivially.
Since your QVTp2QVTg just helps qvtp2qvts it probably migrates to
the qvtp2qvts package.
(The schedule and its dependencies are closely related so I suspect
that splitting them as two models is confusing rather than helpful.)
Regards
Ed Willink
On 18/12/2015 15:13, Adolfo
Sanchez-Barbudo Herrera wrote:
Hi Ed,
I understand what is going on...
Firstly, I left master while go.... but I can manage. I "simply"
have to rebase my commits on ewillink/master. Unfortunately, some
of them include changes to code which doesn't exist anymore, or
perhaps it's somewhere else... Since I have to rely on the "new" I
guess that what I changed before should be irrelevant now.
QVTg (dependency graph) appeared when Horacio factored out QVTs
into scheduler and dependencies distinct meta-models. Now both are
mixed again, so not sure what QVTg "means" now.... A pity that
the good work done on the schedule metamodel refactor has been
thrashed... but anyway, not my concern.
I just want to rebase my commits and try to make all the test
cases work as I had before, but now working on the rebased
code.... let's see
I'll be around trying to come up with something useful...
[P.S: I said the other day, but thanks again for making this new
code appear... I'll try to usefully consume it]
Regards,
Adolfo.
On 18/12/2015 14:56, Ed Willink
wrote:
Hi
Give me a Skype call at 16:00. I suspect you are doing something
irrelevant, or I don't understand something important.
Everything downstream from your CS2AS preparations should be
very similar API-wise. Compare the "new" and not-"New" tests in
OCL2QVTiTestCases.
The not-"New" have no significant changes wrt (M3) master and so
continue to use the Epsilon-based scheduler.
The "New" tests are very similar but use the new scheduler and
QVTp2QVTg.
Regards
Ed Willink
On 18/12/2015 14:27, Adolfo
Sanchez-Barbudo Herrera wrote:
I didn't say master, did I ? I understand there is not point
to start a discussion on this, since it will probably be a
waste of time... I'll see what to do. Among other things,
renaming the QVTp2QVTg java class to QVTp2QVTs ;).
Aborting the merge and taking another approach (interative
rebase)...
Regards,
Adolfo.
On 18/12/2015 13:37, Ed Willink
wrote:
HI
AFAIK nothing has disappeared from master. The dependencies
plugin was part of the refactoring for the old Java-based
scheduler that never passed all tests and so was never
promoted to master.
Regards
Ed Willink
On 18/12/2015 13:15, Adolfo
Sanchez-Barbudo Herrera wrote:
Hi Ed,
This has been turn out to be quite tough. Let's start with
a simple question: Any particular reason about why
dependencies plugin has disappeared ?. There was anything
wrong with the agreed refactoring of splitting QVTs into
QVTg and QVTs ?
Regards,
Adolfo.
On 18/12/2015 11:03, Ed
Willink wrote:
Hi
On 17/12/2015 08:08, Ed
Willink wrote:
I've only had time to promote the 'neutral half' of
ewillink/480567 and I'm not totally confident that
it's a coherent half. Please wait till you see the
repackaged commits for the other helf before
assessing/merging.
Almost all the rest is now on ewillink/master, with
tests for the new currently not replacement scheduler.
The New UpperToLower tests fail intermittently; WIP.
The support for MappingCallBinding.isPolled is currently
retained; it was deleted on ewillink/480567.
The attempted improvements in the interpreted evaluator
when waiting for oclContainer relationships have not
been pushed. They solved a couple of problems but mainly
highlighted that the much better support for the CG
evaluator had not been backported to the interpreter.
Rather than making further attempts to improve the
interpreter, I'll try to ensure that the interpreter
shares the same run-time support code as the CG
evaluator.
Regards
Ed Willink
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
_______________________________________________
qvtd-dev mailing list
qvtd-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/qvtd-dev
|
