Home » Modeling » OCL » ASTNode - transient positions
|
Re: ASTNode - transient positions [message #58314 is a reply to message #58262] |
Wed, 11 June 2008 13:29 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: cdamus.zeligsoft.com
--=-9YOtbIGUPXoobDW+BtA+
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Hi, Jonathan,
That "other model" you speak of could well be the CST model from when
you originally parsed the text. That would be my recommendation,
because if and when MDT OCL needs to go to version 2.0 because of
breaking API changes in a new version of the OCL specification, I will
probably remove the ASTNode interface and these transient properties
from the AST, to maintain strict compliance with OCL. Since the CST was
published as API, these transient properties really are obsolete.
Cheers,
Christian
On Wed, 2008-06-11 at 13:45 +0200, Jonathan MUSSET wrote:
> Christian,
>
> I really have a problem with the ASTNode positions.
> I mean...
> The MTL abstract syntax is based on the OCL Ecore metamodel.
> The problem is that the entry point of code generation is the AST, and
> not the CST.
> At the runtime step, I have only the AST, I don't want to parse the MTL
> file another time.
> How can I do for the debug mode. I need to find the good position, but I
> have only the AST.
> At the moment, the only way for me is to parse the template each time I
> want to generate the code.
> Another solution is to create another model next to my AST model with
> the positions.
>
> Cheers,
>
> Jonathan
--=-9YOtbIGUPXoobDW+BtA+
Content-Type: text/html; charset=utf-8
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.16.0">
</HEAD>
<BODY>
Hi, Jonathan,<BR>
<BR>
That "other model" you speak of could well be the CST model from when you originally parsed the text. That would be my recommendation, because if and when MDT OCL needs to go to version 2.0 because of breaking API changes in a new version of the OCL specification, I will probably remove the ASTNode interface and these transient properties from the AST, to maintain strict compliance with OCL. Since the CST was published as API, these transient properties really are obsolete.<BR>
<BR>
Cheers,<BR>
<BR>
Christian<BR>
<BR>
On Wed, 2008-06-11 at 13:45 +0200, Jonathan MUSSET wrote:
<BLOCKQUOTE TYPE=CITE>
<PRE>
<FONT COLOR="#000000">Christian,</FONT>
<FONT COLOR="#000000">I really have a problem with the ASTNode positions.</FONT>
<FONT COLOR="#000000">I mean...</FONT>
<FONT COLOR="#000000">The MTL abstract syntax is based on the OCL Ecore metamodel.</FONT>
<FONT COLOR="#000000">The problem is that the entry point of code generation is the AST, and </FONT>
<FONT COLOR="#000000">not the CST.</FONT>
<FONT COLOR="#000000">At the runtime step, I have only the AST, I don't want to parse the MTL </FONT>
<FONT COLOR="#000000">file another time.</FONT>
<FONT COLOR="#000000">How can I do for the debug mode. I need to find the good position, but I </FONT>
<FONT COLOR="#000000">have only the AST.</FONT>
<FONT COLOR="#000000">At the moment, the only way for me is to parse the template each time I </FONT>
<FONT COLOR="#000000">want to generate the code.</FONT>
<FONT COLOR="#000000">Another solution is to create another model next to my AST model with </FONT>
<FONT COLOR="#000000">the positions.</FONT>
<FONT COLOR="#000000">Cheers,</FONT>
<FONT COLOR="#000000">Jonathan</FONT>
</PRE>
</BLOCKQUOTE>
</BODY>
</HTML>
--=-9YOtbIGUPXoobDW+BtA+--
|
|
| |
Re: ASTNode - transient positions [message #58383 is a reply to message #58314] |
Wed, 11 June 2008 15:58 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Jonathan MUSSET![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=jonathan.musset%40obeo.fr) Messages: 43 Registered: July 2009 |
Member |
|
|
You're right
I can serialize my CST and my AST model, but it's strange.
Only the position is missing in my AST... :-(
Let's take an example :
- I want to visit the AST to generate the code
- If I have an exception, I want to display the line of the problem
(for example, in java, the line number of an exception is given, where
is the information?)
- If I serialize the CST model only, I have to create my AST each time I
need it... damage! ;-)
- Why not asking the OMG/OCL guys to add that kind of information in
their metamodel?
Cheers
Jonathan
Christian W. Damus a écrit :
> Hi, Jonathan,
>
> That "other model" you speak of could well be the CST model from when
> you originally parsed the text. That would be my recommendation,
> because if and when MDT OCL needs to go to version 2.0 because of
> breaking API changes in a new version of the OCL specification, I will
> probably remove the ASTNode interface and these transient properties
> from the AST, to maintain strict compliance with OCL. Since the CST was
> published as API, these transient properties really are obsolete.
>
> Cheers,
>
> Christian
>
> On Wed, 2008-06-11 at 13:45 +0200, Jonathan MUSSET wrote:
>> Christian,
>>
>> I really have a problem with the ASTNode positions.
>> I mean...
>> The MTL abstract syntax is based on the OCL Ecore metamodel.
>> The problem is that the entry point of code generation is the AST, and
>> not the CST.
>> At the runtime step, I have only the AST, I don't want to parse the MTL
>> file another time.
>> How can I do for the debug mode. I need to find the good position, but I
>> have only the AST.
>> At the moment, the only way for me is to parse the template each time I
>> want to generate the code.
>> Another solution is to create another model next to my AST model with
>> the positions.
>>
>> Cheers,
>>
>> Jonathan
|
|
|
Re: ASTNode - transient positions [message #58423 is a reply to message #58360] |
Wed, 11 June 2008 22:27 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: cdamus.zeligsoft.com
--=-quzT4ja6YQYxEdv0PIJw
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Hi, Adolfo,
Visitable is crucial to the visitor pattern and it only defines
operations, anyway, which have no impact on serialization. I doubt it
would go anywhere. PredefinedType could possibly be obsoleted by a
better implementation of the OCL Standard Library (probably at least
including Ed W.'s suggestion of generating EPackages for the different
metamodel bindings of the library).
I have only just begun turning over ideas for (r)evolution of the API,
and of course the status of the OCL RTF is up in the air, anyway. Any
discussion that we can have, here, will certainly help to shape the
eventual outcome. Don't take anything I say now as definitive.
cW
On Wed, 2008-06-11 at 16:19 +0100, Adolfo S=C3=A1nchez-Barbudo Herrera wrot=
e:
> Hi Christian
>=20
> Just a curiosity:
>=20
> > I will probably remove the ASTNode interface and these transient proper=
ties=20
> > from the AST, to maintain strict compliance with OCL.
>=20
> Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going=20
> out from the metamodel to maintain strict compliance with OMG's OCL ? :)
>=20
> Cheers,
> Adolfo.
--=-quzT4ja6YQYxEdv0PIJw
Content-Type: text/html; charset=utf-8
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.16.0">
</HEAD>
<BODY>
Hi, Adolfo,<BR>
<BR>
Visitable is crucial to the visitor pattern and it only defines operations, anyway, which have no impact on serialization. I doubt it would go anywhere. PredefinedType could possibly be obsoleted by a better implementation of the OCL Standard Library (probably at least including Ed W.'s suggestion of generating EPackages for the different metamodel bindings of the library).<BR>
<BR>
I have only just begun turning over ideas for (r)evolution of the API, and of course the status of the OCL RTF is up in the air, anyway. Any discussion that we can have, here, will certainly help to shape the eventual outcome. Don't take anything I say now as definitive.<BR>
<BR>
cW<BR>
<BR>
On Wed, 2008-06-11 at 16:19 +0100, Adolfo Sánchez-Barbudo Herrera wrote:
<BLOCKQUOTE TYPE=CITE>
<PRE>
<FONT COLOR="#000000">Hi Christian</FONT>
<FONT COLOR="#000000">Just a curiosity:</FONT>
<FONT COLOR="#000000">> I will probably remove the ASTNode interface and these transient properties </FONT>
<FONT COLOR="#000000">> from the AST, to maintain strict compliance with OCL.</FONT>
<FONT COLOR="#000000">Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going </FONT>
<FONT COLOR="#000000">out from the metamodel to maintain strict compliance with OMG's OCL ? :)</FONT>
<FONT COLOR="#000000">Cheers,</FONT>
<FONT COLOR="#000000">Adolfo.</FONT>
</PRE>
</BLOCKQUOTE>
</BODY>
</HTML>
--=-quzT4ja6YQYxEdv0PIJw--
|
|
|
Re: ASTNode - transient positions [message #58548 is a reply to message #58383] |
Wed, 11 June 2008 22:30 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: cdamus.zeligsoft.com
--=-0wHjT68k2xGwr9UyADDW
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Hi, Jonathan,
Yes, traceability to the original source text is valuable for a number
of reasons, including positioning problem markers, refactoring,and more.
However, the Abstract Syntax Model is just that: the *abstract* syntax.
The textual representation really is the business of the concrete syntax
model. I wouldn't want to ask the OMG to mix these two worlds.
Cheers,
Christian
On Wed, 2008-06-11 at 17:58 +0200, Jonathan MUSSET wrote:
> You're right
> I can serialize my CST and my AST model, but it's strange.
> Only the position is missing in my AST... :-(
> Let's take an example :
> - I want to visit the AST to generate the code
> - If I have an exception, I want to display the line of the problem
> (for example, in java, the line number of an exception is given, where=20
> is the information?)
> - If I serialize the CST model only, I have to create my AST each time I=20
> need it... damage! ;-)
> - Why not asking the OMG/OCL guys to add that kind of information in=20
> their metamodel?
>=20
> Cheers
>=20
> Jonathan
>=20
>=20
>=20
> Christian W. Damus a =C3=A9crit :
> > Hi, Jonathan,
> >=20
> > That "other model" you speak of could well be the CST model from when=20
> > you originally parsed the text. That would be my recommendation,=20
> > because if and when MDT OCL needs to go to version 2.0 because of=20
> > breaking API changes in a new version of the OCL specification, I will=20
> > probably remove the ASTNode interface and these transient properties=20
> > from the AST, to maintain strict compliance with OCL. Since the CST wa=
s=20
> > published as API, these transient properties really are obsolete.
> >=20
> > Cheers,
> >=20
> > Christian
> >=20
> > On Wed, 2008-06-11 at 13:45 +0200, Jonathan MUSSET wrote:
> >> Christian,
> >>
> >> I really have a problem with the ASTNode positions.
> >> I mean...
> >> The MTL abstract syntax is based on the OCL Ecore metamodel.
> >> The problem is that the entry point of code generation is the AST, and=
=20
> >> not the CST.
> >> At the runtime step, I have only the AST, I don't want to parse the MT=
L=20
> >> file another time.
> >> How can I do for the debug mode. I need to find the good position, but=
I=20
> >> have only the AST.
> >> At the moment, the only way for me is to parse the template each time =
I=20
> >> want to generate the code.
> >> Another solution is to create another model next to my AST model with=20
> >> the positions.
> >>
> >> Cheers,
> >>
> >> Jonathan
--=-0wHjT68k2xGwr9UyADDW
Content-Type: text/html; charset=utf-8
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.16.0">
</HEAD>
<BODY>
Hi, Jonathan,<BR>
<BR>
Yes, traceability to the original source text is valuable for a number of reasons, including positioning problem markers, refactoring,and more. However, the Abstract Syntax Model is just that: the *abstract* syntax. The textual representation really is the business of the concrete syntax model. I wouldn't want to ask the OMG to mix these two worlds.<BR>
<BR>
Cheers,<BR>
<BR>
Christian<BR>
<BR>
On Wed, 2008-06-11 at 17:58 +0200, Jonathan MUSSET wrote:
<BLOCKQUOTE TYPE=CITE>
<PRE>
<FONT COLOR="#000000">You're right</FONT>
<FONT COLOR="#000000">I can serialize my CST and my AST model, but it's strange.</FONT>
<FONT COLOR="#000000">Only the position is missing in my AST... :-(</FONT>
<FONT COLOR="#000000">Let's take an example :</FONT>
<FONT COLOR="#000000">- I want to visit the AST to generate the code</FONT>
<FONT COLOR="#000000">- If I have an exception, I want to display the line of the problem</FONT>
<FONT COLOR="#000000"> (for example, in java, the line number of an exception is given, where </FONT>
<FONT COLOR="#000000">is the information?)</FONT>
<FONT COLOR="#000000">- If I serialize the CST model only, I have to create my AST each time I </FONT>
<FONT COLOR="#000000">need it... damage! ;-)</FONT>
<FONT COLOR="#000000">- Why not asking the OMG/OCL guys to add that kind of information in </FONT>
<FONT COLOR="#000000">their metamodel?</FONT>
<FONT COLOR="#000000">Cheers</FONT>
<FONT COLOR="#000000">Jonathan</FONT>
<FONT COLOR="#000000">Christian W. Damus a écrit :</FONT>
<FONT COLOR="#000000">> Hi, Jonathan,</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> That "other model" you speak of could well be the CST model from when </FONT>
<FONT COLOR="#000000">> you originally parsed the text. That would be my recommendation, </FONT>
<FONT COLOR="#000000">> because if and when MDT OCL needs to go to version 2.0 because of </FONT>
<FONT COLOR="#000000">> breaking API changes in a new version of the OCL specification, I will </FONT>
<FONT COLOR="#000000">> probably remove the ASTNode interface and these transient properties </FONT>
<FONT COLOR="#000000">> from the AST, to maintain strict compliance with OCL. Since the CST was </FONT>
<FONT COLOR="#000000">> published as API, these transient properties really are obsolete.</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> Cheers,</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> Christian</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> On Wed, 2008-06-11 at 13:45 +0200, Jonathan MUSSET wrote:</FONT>
<FONT COLOR="#000000">>> Christian,</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> I really have a problem with the ASTNode positions.</FONT>
<FONT COLOR="#000000">>> I mean...</FONT>
<FONT COLOR="#000000">>> The MTL abstract syntax is based on the OCL Ecore metamodel.</FONT>
<FONT COLOR="#000000">>> The problem is that the entry point of code generation is the AST, and </FONT>
<FONT COLOR="#000000">>> not the CST.</FONT>
<FONT COLOR="#000000">>> At the runtime step, I have only the AST, I don't want to parse the MTL </FONT>
<FONT COLOR="#000000">>> file another time.</FONT>
<FONT COLOR="#000000">>> How can I do for the debug mode. I need to find the good position, but I </FONT>
<FONT COLOR="#000000">>> have only the AST.</FONT>
<FONT COLOR="#000000">>> At the moment, the only way for me is to parse the template each time I </FONT>
<FONT COLOR="#000000">>> want to generate the code.</FONT>
<FONT COLOR="#000000">>> Another solution is to create another model next to my AST model with </FONT>
<FONT COLOR="#000000">>> the positions.</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> Cheers,</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> Jonathan</FONT>
</PRE>
</BLOCKQUOTE>
</BODY>
</HTML>
--=-0wHjT68k2xGwr9UyADDW--
|
|
|
Re: ASTNode - transient positions [message #58596 is a reply to message #58423] |
Thu, 12 June 2008 12:55 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Adolfo Sanchez-Barbudo Herrera![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=adolfosbh%40gmail.com) Messages: 260 Registered: July 2009 |
Senior Member |
|
|
Uou Revolution !!!
I think that we'll enjoy :).
As you know, here you have a guy who has been "fighting" with MDT-OCL
API from time ago. Although I don't control all the implementation of
it, I have no doubts that if you point me out to a place where those
ideas can be debatable, I will provide, as far as I'm be able to, more
thoughts to them ;).
If you have a bugzilla, wiki, or anything where you are sorting ideas,
please don't hesitate to tell me.
If not, we can create a thread in the newsgroup ;P.
Cheers,
Adolfo.
Christian W. Damus escribió:
> Hi, Adolfo,
>
> Visitable is crucial to the visitor pattern and it only defines
> operations, anyway, which have no impact on serialization. I doubt it
> would go anywhere. PredefinedType could possibly be obsoleted by a
> better implementation of the OCL Standard Library (probably at least
> including Ed W.'s suggestion of generating EPackages for the different
> metamodel bindings of the library).
>
> I have only just begun turning over ideas for (r)evolution of the API,
> and of course the status of the OCL RTF is up in the air, anyway. Any
> discussion that we can have, here, will certainly help to shape the
> eventual outcome. Don't take anything I say now as definitive.
>
> cW
>
> On Wed, 2008-06-11 at 16:19 +0100, Adolfo Sánchez-Barbudo Herrera wrote:
>> Hi Christian
>>
>> Just a curiosity:
>>
>> > I will probably remove the ASTNode interface and these transient properties
>> > from the AST, to maintain strict compliance with OCL.
>>
>> Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going
>> out from the metamodel to maintain strict compliance with OMG's OCL ? :)
>>
>> Cheers,
>> Adolfo.
|
|
|
Re: ASTNode - transient positions [message #58621 is a reply to message #58596] |
Thu, 12 June 2008 21:51 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) ![Go to next message Go to next message](theme/Solstice/images/down.png) |
Eclipse User![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=) |
|
|
|
Originally posted by: cdamus.zeligsoft.com
--=-Bq2RQh/ZAK0Peoh0zNqs
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Hi, Adolfo,
Keep in mind that I am not inclined to revolutionize anything unless an
external factor such as an OCL 2.1 specification that has incompatible
changes forces the issue.
However, it's never too early to start sharing ideas for the future.
Perhaps you would like to raise an enhancement request in bugzilla to
kick-off the discussion?
Cheers,
Christian
On Thu, 2008-06-12 at 13:55 +0100, Adolfo S=C3=A1nchez-Barbudo Herrera wrot=
e:
> Uou Revolution !!!
>=20
> I think that we'll enjoy :).
>=20
> As you know, here you have a guy who has been "fighting" with MDT-OCL=20
> API from time ago. Although I don't control all the implementation of=20
> it, I have no doubts that if you point me out to a place where those=20
> ideas can be debatable, I will provide, as far as I'm be able to, more=20
> thoughts to them ;).
>=20
> If you have a bugzilla, wiki, or anything where you are sorting ideas,=20
> please don't hesitate to tell me.
>=20
> If not, we can create a thread in the newsgroup ;P.
>=20
> Cheers,
> Adolfo.
>=20
> Christian W. Damus escribi=C3=B3:
> > Hi, Adolfo,
> >=20
> > Visitable is crucial to the visitor pattern and it only defines=20
> > operations, anyway, which have no impact on serialization. I doubt it=20
> > would go anywhere. PredefinedType could possibly be obsoleted by a=20
> > better implementation of the OCL Standard Library (probably at least=20
> > including Ed W.'s suggestion of generating EPackages for the different=20
> > metamodel bindings of the library).
> >=20
> > I have only just begun turning over ideas for (r)evolution of the API,=20
> > and of course the status of the OCL RTF is up in the air, anyway. Any=20
> > discussion that we can have, here, will certainly help to shape the=20
> > eventual outcome. Don't take anything I say now as definitive.
> >=20
> > cW
> >=20
> > On Wed, 2008-06-11 at 16:19 +0100, Adolfo S=C3=A1nchez-Barbudo Herrera =
wrote:
> >> Hi Christian
> >>
> >> Just a curiosity:
> >>
> >> > I will probably remove the ASTNode interface and these transient pro=
perties=20
> >> > from the AST, to maintain strict compliance with OCL.
> >>
> >> Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going=20
> >> out from the metamodel to maintain strict compliance with OMG's OCL ? =
:)
> >>
> >> Cheers,
> >> Adolfo.
--=-Bq2RQh/ZAK0Peoh0zNqs
Content-Type: text/html; charset=utf-8
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/3.16.0">
</HEAD>
<BODY>
Hi, Adolfo,<BR>
<BR>
Keep in mind that I am not inclined to revolutionize anything unless an external factor such as an OCL 2.1 specification that has incompatible changes forces the issue.<BR>
<BR>
However, it's never too early to start sharing ideas for the future. Perhaps you would like to raise an enhancement request in bugzilla to kick-off the discussion?<BR>
<BR>
Cheers,<BR>
<BR>
Christian<BR>
<BR>
On Thu, 2008-06-12 at 13:55 +0100, Adolfo Sánchez-Barbudo Herrera wrote:
<BLOCKQUOTE TYPE=CITE>
<PRE>
<FONT COLOR="#000000">Uou Revolution !!!</FONT>
<FONT COLOR="#000000">I think that we'll enjoy :).</FONT>
<FONT COLOR="#000000">As you know, here you have a guy who has been "fighting" with MDT-OCL </FONT>
<FONT COLOR="#000000">API from time ago. Although I don't control all the implementation of </FONT>
<FONT COLOR="#000000">it, I have no doubts that if you point me out to a place where those </FONT>
<FONT COLOR="#000000">ideas can be debatable, I will provide, as far as I'm be able to, more </FONT>
<FONT COLOR="#000000">thoughts to them ;).</FONT>
<FONT COLOR="#000000">If you have a bugzilla, wiki, or anything where you are sorting ideas, </FONT>
<FONT COLOR="#000000">please don't hesitate to tell me.</FONT>
<FONT COLOR="#000000">If not, we can create a thread in the newsgroup ;P.</FONT>
<FONT COLOR="#000000">Cheers,</FONT>
<FONT COLOR="#000000">Adolfo.</FONT>
<FONT COLOR="#000000">Christian W. Damus escribió:</FONT>
<FONT COLOR="#000000">> Hi, Adolfo,</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> Visitable is crucial to the visitor pattern and it only defines </FONT>
<FONT COLOR="#000000">> operations, anyway, which have no impact on serialization. I doubt it </FONT>
<FONT COLOR="#000000">> would go anywhere. PredefinedType could possibly be obsoleted by a </FONT>
<FONT COLOR="#000000">> better implementation of the OCL Standard Library (probably at least </FONT>
<FONT COLOR="#000000">> including Ed W.'s suggestion of generating EPackages for the different </FONT>
<FONT COLOR="#000000">> metamodel bindings of the library).</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> I have only just begun turning over ideas for (r)evolution of the API, </FONT>
<FONT COLOR="#000000">> and of course the status of the OCL RTF is up in the air, anyway. Any </FONT>
<FONT COLOR="#000000">> discussion that we can have, here, will certainly help to shape the </FONT>
<FONT COLOR="#000000">> eventual outcome. Don't take anything I say now as definitive.</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> cW</FONT>
<FONT COLOR="#000000">> </FONT>
<FONT COLOR="#000000">> On Wed, 2008-06-11 at 16:19 +0100, Adolfo Sánchez-Barbudo Herrera wrote:</FONT>
<FONT COLOR="#000000">>> Hi Christian</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> Just a curiosity:</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> > I will probably remove the ASTNode interface and these transient properties </FONT>
<FONT COLOR="#000000">>> > from the AST, to maintain strict compliance with OCL.</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going </FONT>
<FONT COLOR="#000000">>> out from the metamodel to maintain strict compliance with OMG's OCL ? :)</FONT>
<FONT COLOR="#000000">>></FONT>
<FONT COLOR="#000000">>> Cheers,</FONT>
<FONT COLOR="#000000">>> Adolfo.</FONT>
</PRE>
</BLOCKQUOTE>
</BODY>
</HTML>
--=-Bq2RQh/ZAK0Peoh0zNqs--
|
|
|
Re: ASTNode - transient positions [message #58771 is a reply to message #58621] |
Tue, 17 June 2008 08:50 ![Go to previous message Go to previous message](theme/Solstice/images/up.png) |
Adolfo Sanchez-Barbudo Herrera![Friend of Eclipse Friend](/donate/web-api/friends_decorator.php?email=adolfosbh%40gmail.com) Messages: 260 Registered: July 2009 |
Senior Member |
|
|
Hi Christian,
Some in-lined comments below:
> Hi, Adolfo,
>
> Keep in mind that I am not inclined to revolutionize anything unless an
> external factor such as an OCL 2.1 specification that has incompatible
> changes forces the issue.
Yes you are right. But I'm sure that MDT-OCL would gain in clarity,
coherence, extensibility, etc, if some API aspects were restructured.
>
> However, it's never too early to start sharing ideas for the future.
> Perhaps you would like to raise an enhancement request in bugzilla to
> kick-off the discussion?
Sure, we can go on with the discussion there.
https://bugs.eclipse.org/bugs/show_bug.cgi?id=237438
>
> Cheers,
>
> Christian
>
> On Thu, 2008-06-12 at 13:55 +0100, Adolfo Sánchez-Barbudo Herrera wrote:
>> Uou Revolution !!!
>>
>> I think that we'll enjoy :).
>>
>> As you know, here you have a guy who has been "fighting" with MDT-OCL
>> API from time ago. Although I don't control all the implementation of
>> it, I have no doubts that if you point me out to a place where those
>> ideas can be debatable, I will provide, as far as I'm be able to, more
>> thoughts to them ;).
>>
>> If you have a bugzilla, wiki, or anything where you are sorting ideas,
>> please don't hesitate to tell me.
>>
>> If not, we can create a thread in the newsgroup ;P.
>>
>> Cheers,
>> Adolfo.
>>
>> Christian W. Damus escribió:
>> > Hi, Adolfo,
>> >
>> > Visitable is crucial to the visitor pattern and it only defines
>> > operations, anyway, which have no impact on serialization. I doubt it
>> > would go anywhere. PredefinedType could possibly be obsoleted by a
>> > better implementation of the OCL Standard Library (probably at least
>> > including Ed W.'s suggestion of generating EPackages for the different
>> > metamodel bindings of the library).
>> >
>> > I have only just begun turning over ideas for (r)evolution of the API,
>> > and of course the status of the OCL RTF is up in the air, anyway. Any
>> > discussion that we can have, here, will certainly help to shape the
>> > eventual outcome. Don't take anything I say now as definitive.
>> >
>> > cW
>> >
>> > On Wed, 2008-06-11 at 16:19 +0100, Adolfo Sánchez-Barbudo Herrera wrote:
>> >> Hi Christian
>> >>
>> >> Just a curiosity:
>> >>
>> >> > I will probably remove the ASTNode interface and these transient properties
>> >> > from the AST, to maintain strict compliance with OCL.
>> >>
>> >> Are all the non-ocl interfaces (Visitable, PredefinedType, etc) going
>> >> out from the metamodel to maintain strict compliance with OMG's OCL ? :)
>> >>
>> >> Cheers,
>> >> Adolfo.
|
|
|
Goto Forum:
Current Time: Fri Aug 16 16:50:08 GMT 2024
Powered by FUDForum. Page generated in 0.03430 seconds
|