Hi Ed,
Comments in-lined below:
Ed Willink escribió:
Hi Adolfo
QVTo is tightly coupled like QVTd to the OCL parser, and QVTo has very
limited committer resources so we must supply a complete (including
generated files) patch so that QVTo can just Apply Patch and go. (I'll
do QVTd). When you're ready, please raise a QVTo bug for the patch
blocked by the OCL bug. Since QVTo must catch up, the two phases (LPG2
and refactoring) must appear as a single change to HEAD. So we will
need the branch for review, but at least there should now not be
concurrent parser changes. However the Token package import is in
numerous places, so it may be a good idea to the refactoring first so
that it can be reviewed, the migration has already been +1'd subject to
Orbit.
What about M2T Acceleo project which also has to refactor their
grammars ?. I don't don't have any problem in working on QVTo side.
Laurent could you ask your colleagues about this point? In any case,
I'll send to the correspondent proejct dev lists a message when all the
migration committed.
When I started the migration work, I created a patch with grammars
migration and parser refactoring which had to be reworked and split
into two phases. So I would prefer to keep this idea, from any
perspective: OCL and each depending languages (QVTo, Mof M2T, etc). The
problem now as you say is the time, I don't really know when LPG v2
will be IP approved (I'll look into this), but my idea is:
1. Create a patch against HEAD with phase 1, so we ensure as soon as
possible that we and depending languages may work with LPGv2.
2. Create a patch against HEAD with phase 1 for QVTo, to demonstrate
point 1 and to help QVTo team.
3. Commit patches 1 and 2 (chekpoint A). Now we have succesfully
migrated to LPGv2.
4. Create a patch against HEAD with phase 2, we could investigate a
possible refactoring of
5. Create a patch agains HEAD with phase 2 for QVTo., we could ensure
that the refactoring
6. Commit patches 4 and 5 (checkpoint B). Now we have also refactored
the parsing infrastructure.
I would be very worried if we don't reach checkpoint A for M5.
I would be very happy reaching checkpoint B, but I won't be worried if
we aren't able to do it.
If grammars and parsers are not modified we could do 1,2, even 4 and 5,
so the refactoring (phase 2) could be reviewed meanwhile.
If after a couple of weeks, we haven't successfully made LPGv2 be
included into Orbit due to IP approval delays, I'll surely align
patches 1 and 2 to any parallel grammar/parser change. We will take a
decision about what to do with the patches 4 and 5 (Commit patches to
the branch, giving up any refactoring, etc).
I can concentrate on the evaluator, registry and QVTd builds for a
couple of weeks, but I cannot leave the analyzer and environments for
long. It would help to have an estimate timescale for
a) LPG 2 in Orbit
b) LPG 2 usage visible on branch
c) LPG 2 refactoring visible on branch
If LPG 2 is 'guaranteed' to be in Orbit for M5 , OCL and QVTo build
maps can temporarily reference SourceForge.
The project of the Orbit's bundle is not the same of the SourceForge
project's, so I'm not sure if this could be possible. Besides, I'm not
sure if it's good idea an Eclipse (even temporally) build refers to a
non-Eclipse repository, actually I have been advised to not creating
the project into Orbit's repository until IP approval is granted.
Cheers,
Adolfo.
Ed
Adolfo Sánchez-Barbudo Herrera wrote:
Hi
Team,
The LPGv2.0.17 migration (phase 1) is nearly completed:
- The most important issue is pending: IP approval. It's supposed to be
started by Robert Fuhrer today (I have delegated this task to him since
he has some experience with previous LPGv2.0.x CQs).
- I have migrated the current state of the MDT-OCL grammars to
acomodate LPGv2. I did it from scratch (following some guidelines I
made) since a lot of grammar changes have been made after I did the
migration in the experimental branch.
- I have created the new Orbit bundle (in my workspace) and some other
files modification so that LPGv2 is integrated into Orbit. The new
bundle will be called net.sourceforge.lpg.runtime.java
- All this stuff will be committed as soon as IP approval is granted
and when M4 is released.
- I have created a simple migration guide for extending languages. I
hope to improve it when the phase2 is done. http://wiki.eclipse.org/MDT/OCL/MDT-OCL_3.0.0_Migration_Guide#LPG_v2_migration
Regards,
Adolfo.
--
|
Adolfo
Sánchez-Barbudo Herrera
adolfosbh(at)opencanarias(dot)com
C/Elías Ramos González, 4, ofc. 304
38001 SANTA CRUZ DE TENERIFE
Tel.: +34 922 240231 / +34 617 718268 |
_______________________________________________
mdt-ocl.dev mailing list
mdt-ocl.dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mdt-ocl.dev
_______________________________________________
mdt-ocl.dev mailing list
mdt-ocl.dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/mdt-ocl.dev
--
|
Adolfo
Sánchez-Barbudo Herrera
adolfosbh(at)opencanarias(dot)com
C/Elías Ramos González, 4, ofc. 304
38001 SANTA CRUZ DE TENERIFE
Tel.: +34 922 240231 / +34 617 718268 |
|
