[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
RE: [modeling-pmc] Contribution: template engine for GMF
|
Title: Re: [modeling-pmc] Contribution: template engine for GMF
Rich,
1. oaw guys are in cc of this thread and could
express their objections, if any
2. Definitely. Nobody wants to keep the code when
someone else is ready to take responsibility for it ;)
3. It adds one plug-in with template engine to the
SDK. We don't need to bundle tools (like editor, which is btw much better than
one we use for JET) and tests for the engine with GMF.
Ed,
If we get a ready to use bundle in 8-10 weeks - that
would work. Sorry, I don't belive that would be the case ;) Even if project
creation happens in two months, initial contributions won't be as
straightforward as repackaging. They'll need to solve a lot of issues, provided
there are few contributions, different in nature but within single home -
e.g. what should be kept, what should be combined/redesigned/etc. Dealing with
3rd party dependencies might add few more weeks. It was easy to me - GMF
was the only client, no need to answer questions like "is it general enough to
suit needs of most m2t clients", which would have to be answered with M2T
contributions. Therefore, I feel there won't be engine
ready in 2 months (sorry, better be pessimistic here), rather
3-4, which is M4-M5 - bit too late.
Best wishes,
Artem Tikhomirov
The sandbox approach is not an option at this time
(per the Board of Directors), afaiu.
Im not opposed to temporarily
housing this derivative of xPand within an GMF internal package of the tooling
component, as long as:
- the
oAW team has no objections, as its based on their work
- we
migrate to M2T version asap (Artem to work with oAW/M2T on contributing his
modifications)
- it
has minimal impact on the build/packaging/usage of
GMF
Thanks,
Rich
On 8/22/06 1:02 PM,
"Artem Tikhomirov" <Artem.Tikhomirov@xxxxxxxxxxx>
wrote:
Ed,
Let me question the 'proper process' term :) What's wrong
with GMF having template engine tailored for its own needs? If there was an
engine/project already and I'd try to parallel it, that might be wrong.
There's no project yet, nor even the proposal is finished, let alone
deliverables. To me, it sounds like few months. Few months mean no chances
for GMF 2.0 to use it - there are certain changes we can't afford to make in
M6, for example. Delaying use of technology for a year just because later
there *may* appear 'official' implementation doesn't sound as justification
for 'improper'. Moreover, switch to the 'official' implementation is the
goal; intention of the contribution is just to make it happen in 2.0
timeframe.
GMF team is not going to dedicate its efforts to work on
template engine, it's a mere tool for the team to solve some problems more
efficiently. The engine is mature enough for us to just use
it.
Sandboxing activities might sound as a
solution, though I doubt can be done without severely affecting ongoing GMF
work for (again) few months.
Reusing what is there in GMT is
not an option because of ANTLR dependency and few other third-party
libraries (e.g. code formatting)
Best wishes,
Artem Tikhomirov
From: modeling-pmc-bounces@xxxxxxxxxxx
[mailto:modeling-pmc-bounces@xxxxxxxxxxx]
On Behalf Of Ed Merks
Sent: Tuesday, August 22,
2006 8:26 PM
To: PMC members mailing list
Cc:
voelter@xxxxxxx; Sven Efftinge; modeling-pmc@xxxxxxxxxxx;
modeling-pmc-bounces@xxxxxxxxxxx
Subject: Re:
[modeling-pmc] Contribution: template engine for GMF
Artem,
Being
terminally impatient, I can certainly relate to your desire to take
action now and not wait around for a bunch of administrative delays.
But that being said, the proper process here really is to get
this M2T proposal approved and then do this work within the bounds
of that project. Also keep in mind that any use of LPG
by any project still needs to be legal approved even though OCL is
already approved to use it; of course this will go much faster since the
due diligence is already complete. Could your activities take
place in a sandbox (i.e., not in public CVS) until this M2T thing is
approved? Or could you simply reuse what's in GMT now until a
syntax equivalent version of it is available from M2T?
Sven and Paul,
Do you guys have an ETA
for when the M2T proposal will be ready to submit to the EMO?
(The MDT proposal has already been submitted.)
Ed
Merks/Toronto/IBM@IBMCA
mailto: merks@xxxxxxxxxx
905-413-3265
(t/l
969)
"Artem Tikhomirov"
<Artem.Tikhomirov@xxxxxxxxxxx>
Sent by:
modeling-pmc-bounces@xxxxxxxxxxx 08/22/2006 11:07 AM
Please respond to
PMC members mailing
list <modeling-pmc@xxxxxxxxxxx>
To
<modeling-pmc@xxxxxxxxxxx>
cc
Sven Efftinge <sven@xxxxxxxxxxx>,
voelter@xxxxxxx
Subject
[modeling-pmc] Contribution: template engine
for GMF
Hello,
I'm
writing to inform the PMC about intent to provide significant code
contribution to one of Modeling projects, namely GMF. The
contribution is template engine, to be used along with JET engine,
which is the only template engine used in GMF now.
This new
template engine is based on Xpand template engine which is part of
openArchitectureWare framework, GMT project (http://www.eclipse.org/gmt/oaw).
It's stripped-down version of the framework, with pieces relevant
and essential only for code generation left. All dependencies from
3rd party libraries were removed and the only one left -
parser/grammar framework (ANTLR 2.7), has been replaced with LPG
library (one EMFT uses for OCL). Rest of the original code is
licensed under EPL.
Xpand approach to code generation is different
from JET's and proved to be much more convenient for some tasks we
face with GMF templates. Also, one of initial plan items for GMF
states support for flexible method of generation
(https://bugs.eclipse.org/bugs/show_bug.cgi?id=114207), thus new
engine helps both GMF developers and GMF users to express their
'textual' intentions with a language they feel most
suitable.
oAW framework and Xpand in particular are being
considered as initial contributions to Model to Text project. Also,
M2T project might have support to switch template engines. However,
project set up and first deliveries are points in distant future
(months), and that just means "no" for GMF 2.0. Using GMF-owned
component allows us to start using new engine soon. Syntax of the
new engine is almost identical to that of original, and there won't
be a problem to switch to M2T deliveries once they are out. Note,
however, that proposed contribution in not kind of 'throw away'
stuff, pieces of it might find their way into M2T contributions
(e.g. LPG grammars, test cases, patches and
improvements)
Size of the contribution is about 10KLOC of
handwritten code plus few grammar definitions and LPG generated
code. Contribution is thoroughly covered with unit tests (200+).
I'll need PMC approval to proceed with contribution
questionnaire (http://www.eclipse.org/legal/EclipseLegalProcessPoster.pdf
and http://www.eclipse.org/legal/ContributionQuestionnairePart1-v1.0.php).
Please let me know your opinion and feel free to ask any question
about the contribution. Thank you!
Best wishes,
Artem
Tikhomirov
_______________________________________________
modeling-pmc
mailing
list
modeling-pmc@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/modeling-pmc
_______________________________________________
modeling-pmc
mailing
list
modeling-pmc@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/modeling-pmc
--
Richard
C. Gronback
Borland Software
Corporation
richard.gronback@xxxxxxxxxxx
+1 860 227
9215