Eclipse Community Forums: Modeling (top-level project) » A couple of thoughts on overall strategy

Help

Home

Home » Modeling » Modeling (top-level project) » A couple of thoughts on overall strategy

Show: Today's Messages :: Show Polls :: Message Navigator

A couple of thoughts on overall strategy [message #378412]

Mon, 18 December 2006 19:27

Markus Voelter

Messages: 33
Registered: July 2009

Member

Hi all,

I am currently reviewing a draft book on the Microsoft
DSL tools (written by the appropriate suspects at MS).

Reading this book made me thinking in several ways:

* first of all, if we only compare GMF to the
"frontend" of the DSL tools (and forget about
all the code gen stuff) then I think the MS tool
chain is much easier to learn and grasp than GMF.
This might be due to the really nice tool support.

* I also think that their tooling is not much
less powerful than the Eclipse GMF world, at
least wrt. to the 80% case.

* Considering that the current state is 1.0, and
considering the typically rapid pace at which the
MS folks tend to release newer versions, it is -
in my view - a question of time until they've
caught up with the Eclipse world. I also think
their meta meta model is more powerful than EMF,
and they plan do develop it even further.

* the book provides an excellent, comprehensive
and consistent approach to using the tools -
from frontend to code!

So why am I writing this?

I really think that
a) the Eclipse community needs a bit more
focus on usability and accessibility.
b) we need much more consistency
c) we need a book (no I am not voluteering to
write it :-))

As I mentioned (I think) at the Eclipse Summit, I
really think that most critical role the EMP project
has to play is to care about these three points.

I would like to initiate a discussion with all of you
on how we can move forward in this respect. I think this
is really important.

What do you think?

Cheers,
Markus

--
Markus Völter

voelter - ingenieurbüro für softwaretechnologie
Ziegeläcker 11, 89520 Heidenheim, Germany
Tel. +49 (0) 171 / 86 01 869
Email: voelter@acm.org

Web: http://www.voelter.de
Blog: http://www.voelter.de/blog
Podcast: http://www.se-radio.net

PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378417 is a reply to message #378412]

Tue, 19 December 2006 07:33

Eclipse User

Originally posted by: bezivin.univ-nantes.fr

dear Markus,

I do share your concern.
We have the same readings and, like you, I am much impressed
by the rapid progresses that Steve Cooke, Stuart Kent and their colleagues
have been making in recent months at Microsoft on Model Engineering.

Some time ago we had been building tools to bridge the MS DSL Tools
environment
and the Eclipse modeling environment (Ecore). The recent release in
september 2006 of version V1
of their tools is much more stable, flexible and powerful than the version
we used at that time.
I agree with you that the official publication of their book is going to be
an important step,
as will probably be the next announcements that will be made before summer
2007.

So there exists currently a wealthy competition between the Eclipse Modeling
Project and
the Microsoft DSL Tools intiative and this is not bad. I concur with your
conclusions that the MS DSL Tools
currently have an advantage on integration, usability, and accessibility
over EMP more fragmented facilities.

The good news when reading the forthcoming literature is that both
approaches are based
on common model engineering principles. It's a competition about
implementation and usability
but both approaches are built on similar conceptual architectures, which may
facilitate at a later
time interoperability between them.

This being said, the MS DSL tools take much more seriously than EMP the core
idea of
DSLs (Domain Specific Languages). This may also be one explanation of their
rapid progresses.
We need to work hard in EMP on the idea of providing a practical framework
to easily build
new DSLs by giving abstract syntaxes, visual concrete syntaxes, textual
concrete syntaxes and
semantics.

We have now excellent basic tools like GMF. We need similar facilities for
defining textual
concrete syntaxes. But above all we need global tools to integrate all the
fragmented
facilities available now or later in EMP. We need to put this requirement as
high as possible
in the EMP priority list.

When defining the EMP roadmap, I agree that we should focus on
usability, accessibility, pluggability, composability, and consistency.

In this respect, the following points seems important to me:
a) identifying essential missing facilities.
(avoiding exotic features, looking for really useful functionalities
like "model diff", etc.)
b) providing a practical way to bind various tools together.
(i.e. to orchestrate a model processing chain)
c) start thinking about public libraries of contributed extensible and
reusable DSLs.

Thanks Markus for initiating this useful discussion thread.

Jean

"Markus Voelter" <schogglad@web.de> a

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378420 is a reply to message #378412]

Tue, 19 December 2006 13:49

Richard Gronback

Messages: 605
Registered: July 2009

Senior Member

Hi Markus,

Thanks for starting the discussion. Assuming most agree we are indeed
competing with Visual Studio, I doubt many will disagree with your
assessment and recommendations, which brings us to "how?" question.

Comparing features and capabilities of Visual Studio with that of the
Eclipse Modeling Project is great, but how do we effect the change you
propose? The two "products" are developed quite differently, as you know.

According to Bjorn, "Eclipse is You"
http://eclipse-projects.blogspot.com/2006/12/eclipse-is-you. html, which
implies it's up to you and the rest of the community to contribute what is
needed to make Eclipse (EMP) what you want it to be. That said, if projects
within EMP are guilty of not committing contributed patches, as Ian points
out
( http://irbull.blogspot.com/2006/12/eclipse-is-you-but-some-o f-you-are.html)
, please let us know.

As Doug points out
( http://cdtdoug.blogspot.com/2006/12/eclipse-is-you-and-heres -why.html),
it's not possible for the PMC to make its contributors work on particular
features, as they are (generally) employed by member companies who use
Eclipse as the basis of their commercial products. These products must have
commercial differentiators in order to be viable, which often means adding
those features you mention below. So, perhaps it's the commercial products
you should compare with Visual Studio? ;-)

All this does not mean we can't make meaningful progress within EMP. For
starters, I've added this topic to our PMC call agenda for later this
morning.

Looking at each of your items:

A) We can add usability and accessibility to the EMP project plan for the
next cycle. This cycle, we are focused on restructuring and unification,
which seems a necessary first step.

A number of bugs entered/identified to be worked with 'usability' and even
'helpwanted' keywords would be helpful (with patches, even better ;-). This
gives the issues some degree of visibility, but still doesn't solve the
contribution problem. Any ideas on that one would be welcome. I've
considered several times a "Help Wanted" campaign that would make it much
easier for outside contributors to know what's needed, and be given some
direction on how to most effectively contribute. But, I'm still unsure of
its possibility for success, and frankly haven't found the time to begin.

B) For consistency, the same applies.

C) Regarding the book idea, I think it's critical and would be willing to
contribute for sure (as would others, I suspect). Personally, I've been
waiting on some of the pieces to fall more into place (e.g. QVT
contribution, M2T to become official, UML2 Tools to mature a bit, and now
the latest workflow proposal, etc.). Who else is serious about contributing
to this effort?

Best,
Rich

On 12/18/06 2:27 PM, in article op.tkrscdjt474qjy@markust43, "Markus
Voelter" <schogglad@web.de> wrote:

> Hi all,
>
> I am currently reviewing a draft book on the Microsoft
> DSL tools (written by the appropriate suspects at MS).
>
> Reading this book made me thinking in several ways:
>
> * first of all, if we only compare GMF to the
> "frontend" of the DSL tools (and forget about
> all the code gen stuff) then I think the MS tool
> chain is much easier to learn and grasp than GMF.
> This might be due to the really nice tool support.
>
> * I also think that their tooling is not much
> less powerful than the Eclipse GMF world, at
> least wrt. to the 80% case.
>
> * Considering that the current state is 1.0, and
> considering the typically rapid pace at which the
> MS folks tend to release newer versions, it is -
> in my view - a question of time until they've
> caught up with the Eclipse world. I also think
> their meta meta model is more powerful than EMF,
> and they plan do develop it even further.
>
> * the book provides an excellent, comprehensive
> and consistent approach to using the tools -
> from frontend to code!
>
> So why am I writing this?
>
> I really think that
> a) the Eclipse community needs a bit more
> focus on usability and accessibility.
> b) we need much more consistency
> c) we need a book (no I am not voluteering to
> write it :-))
>
> As I mentioned (I think) at the Eclipse Summit, I
> really think that most critical role the EMP project
> has to play is to care about these three points.
>
> I would like to initiate a discussion with all of you
> on how we can move forward in this respect. I think this
> is really important.
>
> What do you think?
>
> Cheers,
> Markus

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378427 is a reply to message #378412]

Tue, 19 December 2006 19:08

Eclipse User

Originally posted by: koester.matthias.gmx.de

Hi Markus,

I'm working in the modeling area for quite a while (approx. 5 years
now), but I took us sometime to build a CASE tool with the eclipse
frameworks (I'm talking about Apollo for UML from Gentleware AG). The
newsgroups are a great place to learn a lot, but one needs some
experience to ask the right questions;-). A good book would help a lot
to spread the word about EMP.

I think that especially the possibility to generate tools for XML\XSD
based models is very powerful, but isn't very easy to understand. I'm
currently working for a company that build their own persistence layer
for a XML based model format and they spend a lot of time to rebuild
sth. that perhaps is already possible with the standard EMF! A lot of
people now are using different XML binding frameworks, but could benefit
from EMFT and GMF. So I think that this could be one way to show the
power of EMP to people from outside the MDSD community. My impression is
that already a lot of people from the MDSD community are aware of the
eclipse ecosystem, so it would be important to address people without a
modeling background with such a book.

I hope that EMP will lead to more usable tools, since a lot of other
projects (TOPCASED, Papyrus) are planning to use more (or are already
using) components of EMP. But collaboration is essential to gain more
momentum.

Regards,
Matthias

Markus Voelter schrieb:
> Hi all,
>
> I am currently reviewing a draft book on the Microsoft
> DSL tools (written by the appropriate suspects at MS).
>
> Reading this book made me thinking in several ways:
>
> * first of all, if we only compare GMF to the
> "frontend" of the DSL tools (and forget about
> all the code gen stuff) then I think the MS tool
> chain is much easier to learn and grasp than GMF.
> This might be due to the really nice tool support.
>
> * I also think that their tooling is not much
> less powerful than the Eclipse GMF world, at
> least wrt. to the 80% case.
>
> * Considering that the current state is 1.0, and
> considering the typically rapid pace at which the
> MS folks tend to release newer versions, it is -
> in my view - a question of time until they've
> caught up with the Eclipse world. I also think
> their meta meta model is more powerful than EMF,
> and they plan do develop it even further.
>
> * the book provides an excellent, comprehensive
> and consistent approach to using the tools -
> from frontend to code!
>
> So why am I writing this?
>
> I really think that
> a) the Eclipse community needs a bit more
> focus on usability and accessibility.
> b) we need much more consistency
> c) we need a book (no I am not voluteering to
> write it :-))
>
> As I mentioned (I think) at the Eclipse Summit, I
> really think that most critical role the EMP project
> has to play is to care about these three points.
>
> I would like to initiate a discussion with all of you
> on how we can move forward in this respect. I think this
> is really important.
>
> What do you think?
>
> Cheers,
> Markus
>
> --Markus Völter
>
> voelter - ingenieurbüro für softwaretechnologie
> Ziegeläcker 11, 89520 Heidenheim, Germany
> Tel. +49 (0) 171 / 86 01 869
> Email: voelter@acm.org
>
> Web: http://www.voelter.de
> Blog: http://www.voelter.de/blog
> Podcast: http://www.se-radio.net
>
> PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378431 is a reply to message #378417]

Wed, 20 December 2006 17:08

Eclipse User

Originally posted by: jmunoz.dsic.upv.es

Hi all,

Last course we carried out at the Technical University of Valencia an
empirical comparison of DSL Tools and the Eclipse Modeling Plug-ins. In
order to evaluate them, an experiment was developed with 48 last year
undergraduate students of computer science engineering. The students
were divided into 2 groups for developing a DSL (including code
generation) where each group was using a different tool. At the end of
the course, the students answered several questions about their
experiences. The results were presented in a paper that can be found
here (in english):
http://www.dsic.upv.es/workshops/dsdm06/files/dsdm06-11-Pele chano.pdf

Best Regards,

Javier Muñoz

El mar, 19-12-2006 a las 08:33 +0100, Jean.Bezivin escribió:
> dear Markus,
>
> I do share your concern.
> We have the same readings and, like you, I am much impressed
> by the rapid progresses that Steve Cooke, Stuart Kent and their colleagues
> have been making in recent months at Microsoft on Model Engineering.
>
> Some time ago we had been building tools to bridge the MS DSL Tools
> environment
> and the Eclipse modeling environment (Ecore). The recent release in
> september 2006 of version V1
> of their tools is much more stable, flexible and powerful than the version
> we used at that time.
> I agree with you that the official publication of their book is going to be
> an important step,
> as will probably be the next announcements that will be made before summer
> 2007.
>
> So there exists currently a wealthy competition between the Eclipse Modeling
> Project and
> the Microsoft DSL Tools intiative and this is not bad. I concur with your
> conclusions that the MS DSL Tools
> currently have an advantage on integration, usability, and accessibility
> over EMP more fragmented facilities.
>
> The good news when reading the forthcoming literature is that both
> approaches are based
> on common model engineering principles. It's a competition about
> implementation and usability
> but both approaches are built on similar conceptual architectures, which may
> facilitate at a later
> time interoperability between them.
>
> This being said, the MS DSL tools take much more seriously than EMP the core
> idea of
> DSLs (Domain Specific Languages). This may also be one explanation of their
> rapid progresses.
> We need to work hard in EMP on the idea of providing a practical framework
> to easily build
> new DSLs by giving abstract syntaxes, visual concrete syntaxes, textual
> concrete syntaxes and
> semantics.
>
> We have now excellent basic tools like GMF. We need similar facilities for
> defining textual
> concrete syntaxes. But above all we need global tools to integrate all the
> fragmented
> facilities available now or later in EMP. We need to put this requirement as
> high as possible
> in the EMP priority list.
>
> When defining the EMP roadmap, I agree that we should focus on
> usability, accessibility, pluggability, composability, and consistency.
>
> In this respect, the following points seems important to me:
> a) identifying essential missing facilities.
> (avoiding exotic features, looking for really useful functionalities
> like "model diff", etc.)
> b) providing a practical way to bind various tools together.
> (i.e. to orchestrate a model processing chain)
> c) start thinking about public libraries of contributed extensible and
> reusable DSLs.
>
> Thanks Markus for initiating this useful discussion thread.
>
> Jean
>
> "Markus Voelter" <schogglad@web.de> a crit dans le message de news:
> op.tkrscdjt474qjy@markust43...
> > Hi all,
> >
> > I am currently reviewing a draft book on the Microsoft
> > DSL tools (written by the appropriate suspects at MS).
> >
> > Reading this book made me thinking in several ways:
> >
> > * first of all, if we only compare GMF to the
> > "frontend" of the DSL tools (and forget about
> > all the code gen stuff) then I think the MS tool
> > chain is much easier to learn and grasp than GMF.
> > This might be due to the really nice tool support.
> >
> > * I also think that their tooling is not much
> > less powerful than the Eclipse GMF world, at
> > least wrt. to the 80% case.
> >
> > * Considering that the current state is 1.0, and
> > considering the typically rapid pace at which the
> > MS folks tend to release newer versions, it is -
> > in my view - a question of time until they've
> > caught up with the Eclipse world. I also think
> > their meta meta model is more powerful than EMF,
> > and they plan do develop it even further.
> >
> > * the book provides an excellent, comprehensive
> > and consistent approach to using the tools -
> > from frontend to code!
> >
> > So why am I writing this?
> >
> > I really think that
> > a) the Eclipse community needs a bit more
> > focus on usability and accessibility.
> > b) we need much more consistency
> > c) we need a book (no I am not voluteering to
> > write it :-))
> >
> > As I mentioned (I think) at the Eclipse Summit, I
> > really think that most critical role the EMP project
> > has to play is to care about these three points.
> >
> > I would like to initiate a discussion with all of you
> > on how we can move forward in this respect. I think this
> > is really important.
> >
> > What do you think?
> >
> > Cheers,
> > Markus
> >
> > --
> > Markus Vlter
> >
> > voelter - ingenieurbro fr softwaretechnologie
> > Ziegelcker 11, 89520 Heidenheim, Germany
> > Tel. +49 (0) 171 / 86 01 869
> > Email: voelter@acm.org
> >
> > Web: http://www.voelter.de
> > Blog: http://www.voelter.de/blog
> > Podcast: http://www.se-radio.net
> >
> > PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg
>
>

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378435 is a reply to message #378417]

Sun, 24 December 2006 11:46

Markus Voelter

Messages: 33
Registered: July 2009

Member

> The good news when reading the forthcoming literature is that both
> approaches are based on common model engineering principles.

right. That's good for MD(S)D, but doesn't help the
Eclipse community :-)

> This being said, the MS DSL tools take much more seriously than EMP the
> core idea of DSLs (Domain Specific Languages).

right. Although this is in part due to the fact that they
don't like UML - which might also be a disadvantage for them
(although I am much in favour of DSLs, of course).

> We have now excellent basic tools like GMF. We need similar facilities
> for defining textual concrete syntaxes.

yes, very much so!

> When defining the EMP roadmap, I agree that we should focus on
> usability, accessibility, pluggability, composability, and consistency.

very good summary!

Markus

--
Markus Völter

voelter - ingenieurbüro für softwaretechnologie
Ziegeläcker 11, 89520 Heidenheim, Germany
Tel. +49 (0) 171 / 86 01 869
Email: voelter@acm.org

Web: http://www.voelter.de
Blog: http://www.voelter.de/blog
Podcast: http://www.se-radio.net

PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg
--
Markus Völter

voelter - ingenieurbüro für softwaretechnologie
Ziegeläcker 11, 89520 Heidenheim, Germany
Tel. +49 (0) 171 / 86 01 869
Email: voelter@acm.org

Web: http://www.voelter.de
Blog: http://www.voelter.de/blog
Podcast: http://www.se-radio.net

PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg

Report message to a moderator

Re: A couple of thoughts on overall strategy [message #378439 is a reply to message #378420]

Sun, 24 December 2006 11:56

Markus Voelter

Messages: 33
Registered: July 2009

Member

> Thanks for starting the discussion. Assuming most agree we are indeed
> competing with Visual Studio, I doubt many will disagree with your
> assessment and recommendations, which brings us to "how?" question.

right :-)

> Comparing features and capabilities of Visual Studio with that of the
> Eclipse Modeling Project is great, but how do we effect the change you
> propose?

this is exactly what I want to get to as the result of the discussion!

> According to Bjorn, "Eclipse is You"
> http://eclipse-projects.blogspot.com/2006/12/eclipse-is-you. html, which
> implies it's up to you and the rest of the community to contribute what
> is needed to make Eclipse (EMP) what you want it to be. That said, if
> projects within EMP are guilty of not committing contributed patches, as
> Ian points out
> ( http://irbull.blogspot.com/2006/12/eclipse-is-you-but-some-o f-you-are.html)
> , please let us know.

right. Obviously it is about the community. However, that's not all
of the story. It is also about
* how the different projects are coordinated
* which "strategic directions" are set
* which integration steps and documentations are available.

Although the work has to be done by "the community", it is the job
of "the project" to set the priorities and encourage the "right"
subprojects.

> These products must have commercial differentiators in order to be
> viable, which often means adding those features you mention below. So,
> perhaps it's the commercial products you should compare with Visual
> Studio? ;-)

no, not at all! Eclipse is an *Open Source* project, so the Open
Source solution must be viable independent of commercial tools built
on top of it. Otherwise the Open Source thing is just a facade. If the
latter is the idea of Eclipse.org, I cannot identify myself with that.

> All this does not mean we can't make meaningful progress within EMP. For
> starters, I've added this topic to our PMC call agenda for later this
> morning.

ok - and what was the results??

> A) We can add usability and accessibility to the EMP project plan for the
> next cycle. This cycle, we are focused on restructuring and unification,
> which seems a necessary first step.

agreed.

> C) Regarding the book idea, I think it's critical and would be willing to
> contribute for sure (as would others, I suspect). Personally, I've been
> waiting on some of the pieces to fall more into place (e.g. QVT
> contribution, M2T to become official, UML2 Tools to mature a bit, and now
> the latest workflow proposal, etc.). Who else is serious about
> contributing to this effort?

the oAW team certainly is. I mean, the whole point of oAW
has been (and will be even more) to provide a consistent,
usable and integrated tool chain.

Markus

>
> Best,
> Rich
>
> On 12/18/06 2:27 PM, in article op.tkrscdjt474qjy@markust43, "Markus
> Voelter" <schogglad@web.de> wrote:
>
>> Hi all,
>>
>> I am currently reviewing a draft book on the Microsoft
>> DSL tools (written by the appropriate suspects at MS).
>>
>> Reading this book made me thinking in several ways:
>>
>> * first of all, if we only compare GMF to the
>> "frontend" of the DSL tools (and forget about
>> all the code gen stuff) then I think the MS tool
>> chain is much easier to learn and grasp than GMF.
>> This might be due to the really nice tool support.
>>
>> * I also think that their tooling is not much
>> less powerful than the Eclipse GMF world, at
>> least wrt. to the 80% case.
>>
>> * Considering that the current state is 1.0, and
>> considering the typically rapid pace at which the
>> MS folks tend to release newer versions, it is -
>> in my view - a question of time until they've
>> caught up with the Eclipse world. I also think
>> their meta meta model is more powerful than EMF,
>> and they plan do develop it even further.
>>
>> * the book provides an excellent, comprehensive
>> and consistent approach to using the tools -
>> from frontend to code!
>>
>> So why am I writing this?
>>
>> I really think that
>> a) the Eclipse community needs a bit more
>> focus on usability and accessibility.
>> b) we need much more consistency
>> c) we need a book (no I am not voluteering to
>> write it :-))
>>
>> As I mentioned (I think) at the Eclipse Summit, I
>> really think that most critical role the EMP project
>> has to play is to care about these three points.
>>
>> I would like to initiate a discussion with all of you
>> on how we can move forward in this respect. I think this
>> is really important.
>>
>> What do you think?
>>
>> Cheers,
>> Markus
>

--
Markus Völter

voelter - ingenieurbüro für softwaretechnologie
Ziegeläcker 11, 89520 Heidenheim, Germany
Tel. +49 (0) 171 / 86 01 869
Email: voelter@acm.org

Web: http://www.voelter.de
Blog: http://www.voelter.de/blog
Podcast: http://www.se-radio.net

PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg

--
Markus Völter

voelter - ingenieurbüro für softwaretechnologie
Ziegeläcker 11, 89520 Heidenheim, Germany
Tel. +49 (0) 171 / 86 01 869
Email: voelter@acm.org

Web: http://www.voelter.de
Blog: http://www.voelter.de/blog
Podcast: http://www.se-radio.net

PGP Public Key: http://www.voelter.de/data/MarkusVoelter.gpg

Report message to a moderator