Home » Modeling » UML2 » UML templates
| UML templates [message #477131] |
Wed, 26 March 2008 11:19  |
Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Hi,
I have a problem with using templates and don't know if it is caused by
a real issue from the UML spec or if it is something I missed.
A very frequent situation in our component-based modeling process cannot
be properly addressed by the UML templating mechanism as it is now. The
problem we face is similar to the one illustrated in the figure below
and concerns the fact that not always can the same parameter "expose"
(to use the UML lingo) two or more elements.
ClassA<T, n>
prop1: T[n]
prop2: T[n]
The example above works for parameter T, while it fails with n. Yet, the
need of having multiple elements whose multiplicity is parameterized by
a same parameter seems quite plausible. Taking a look at the UML
metamodel we can see that the references between a property and its
type, on one side, and between a property and its upper bound
multiplicity (i.e. upperValue), on the other side, present different
natures --the first case being a simple reference and the second, an
aggregation--, what makes all the difference when you want to expose the
same element multiple times. In our modeling process, most often than
not, a given parameter needs to be used by more than one element inside
a template.
Can anyone comment on this?
Thanks in advance
César
|
|
| |
| Re: UML templates [message #477141 is a reply to message #477139] |
Thu, 27 March 2008 11:32  |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Hi, James,
Thanks for your answer. By the way, I had already read your good article
on generics.
I had also looked for related questions in the UML issues, but to no
avail. In fact what I need to "expose" is not the property
(MultiplicityElement) but its multiplicity (i.e. a ValueSpecification).
And according to the UML spec, value specifications can be exposed. As I
said before, a parameter can expose only one multiplicity value while I
needed to have the same parameter exposing more than one multiplicity
value.
To solve this problem, I use the aggregation type ownedParametedElement
(i.e. the parameterable element belongs to the template parameter). And
this parameterable element, in this case, is my value specification
(say, vs1) that will be used as a value for my exposed multiplicities.
However, when I specify that my property1 has a multiplicity value
(upperValue) equals to vs1, it takes possession of it and I cannot reuse
vs1 as an upperValue for another property (because an element cannot be
owned by more than one "container"). And this contrasts with the
relationship between a property and its type, which is just a simple
reference (that's why in my simple example we can have more than one
property with the same type T)
Schematically, we would have (please make an extra effort to understand
these diagrams ;-)
------------------------
ParameterElement
------------------------
<>
|
v
-----------------------
ParameterableElement
----------------------
<>
|
v
-----------------------
ValueSpecification
----------------------
^
| (aggregation)
<>
------------------------
MultiplicityElement
-----------------------
---------------------------------------------------------
------------------------
ParameterElement
------------------------
<>
|
v
-----------------------
ParameterableElement
----------------------
<>
|
v
-----------------------
Classifier
----------------------
^ ^
| | (simple references)
| |
--------------- --------------
Property1 Property2
--------------- --------------
Hope my problem is better explained now.
César
James Bruck wrote:
> Hi César,
>
> The "exposing" of elements is handled by the TemplateParameter type. From
> the spec "...A TemplateParameter exposes ParameterableElements as formal
> parameters...".
> If I understand your question correctly, you want to specify the
> multiplicity of the property as a parameter. Unfortunately it appears that
> MultiplicityElement is not a ParameterableElement. That would seem to
> imply that one would not be able to specify TemplateParameterSubstitutions
> for some TemplateBinding to modify the bounds of the property.
> I took a quick look at all the RTF issues currently logged and this did not
> appear to be an issue that has been raised.
>
> The following might be of some use...
> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>
> I will dig more deeply into this tomorrow.
>
> Cheers,
> - James.
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:fsdbha$b4i$1@build.eclipse.org...
>> Hi,
>>
>> I have a problem with using templates and don't know if it is caused by a
>> real issue from the UML spec or if it is something I missed.
>>
>> A very frequent situation in our component-based modeling process cannot
>> be properly addressed by the UML templating mechanism as it is now. The
>> problem we face is similar to the one illustrated in the figure below and
>> concerns the fact that not always can the same parameter "expose" (to use
>> the UML lingo) two or more elements.
>>
>> ClassA<T, n>
>> prop1: T[n]
>> prop2: T[n]
>>
>> The example above works for parameter T, while it fails with n. Yet, the
>> need of having multiple elements whose multiplicity is parameterized by a
>> same parameter seems quite plausible. Taking a look at the UML metamodel
>> we can see that the references between a property and its type, on one
>> side, and between a property and its upper bound multiplicity (i.e.
>> upperValue), on the other side, present different natures --the first case
>> being a simple reference and the second, an aggregation--, what makes all
>> the difference when you want to expose the same element multiple times. In
>> our modeling process, most often than not, a given parameter needs to be
>> used by more than one element inside a template.
>>
>> Can anyone comment on this?
>>
>> Thanks in advance
>>
>> César
>
>
|
|
| |
| Re: UML templates [message #477148 is a reply to message #477142] |
Thu, 27 March 2008 16:20 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
> You may want to raise an RTF issue with the OMG on this point.
Is there a reliable way of doing this? Last time I sent a mail to
issues@omg.org, they didn't even bother sending me an automatic reply,
let alone to file the issue.
Thanks for your help anyway.
César
James Bruck wrote:
> Hi César,
>
> The second posting clarified it for me. I see the issue. Unfortunately it
> appears that you will not be able to reuse your value specification.
> As you mentioned, an element cannot be owned by more than one container.
> This seems to be a basic problem that UML should address or at the very
> least it should be brought to the attention of those defining the spec.
>
> You may want to raise an RTF issue with the OMG on this point.
>
> Cheers,
> - James.
>
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:47EB85BA.4050409@yahoo.fr...
>> Hi, James,
>>
>> Thanks for your answer. By the way, I had already read your good article
>> on generics.
>>
>> I had also looked for related questions in the UML issues, but to no
>> avail. In fact what I need to "expose" is not the property
>> (MultiplicityElement) but its multiplicity (i.e. a ValueSpecification).
>> And according to the UML spec, value specifications can be exposed. As I
>> said before, a parameter can expose only one multiplicity value while I
>> needed to have the same parameter exposing more than one multiplicity
>> value.
>>
>> To solve this problem, I use the aggregation type ownedParametedElement
>> (i.e. the parameterable element belongs to the template parameter). And
>> this parameterable element, in this case, is my value specification (say,
>> vs1) that will be used as a value for my exposed multiplicities. However,
>> when I specify that my property1 has a multiplicity value (upperValue)
>> equals to vs1, it takes possession of it and I cannot reuse vs1 as an
>> upperValue for another property (because an element cannot be owned by
>> more than one "container"). And this contrasts with the relationship
>> between a property and its type, which is just a simple reference (that's
>> why in my simple example we can have more than one property with the same
>> type T)
>>
>> Schematically, we would have (please make an extra effort to understand
>> these diagrams ;-)
>>
>> ------------------------
>> ParameterElement
>> ------------------------
>> <>
>> |
>> v
>> -----------------------
>> ParameterableElement
>> ----------------------
>> <>
>> |
>> v
>> -----------------------
>> ValueSpecification
>> ----------------------
>> ^
>> | (aggregation)
>> <>
>> ------------------------
>> MultiplicityElement
>> -----------------------
>>
>> ---------------------------------------------------------
>>
>> ------------------------
>> ParameterElement
>> ------------------------
>> <>
>> |
>> v
>> -----------------------
>> ParameterableElement
>> ----------------------
>> <>
>> |
>> v
>> -----------------------
>> Classifier
>> ----------------------
>> ^ ^
>> | | (simple references)
>> | |
>> --------------- --------------
>> Property1 Property2
>> --------------- --------------
>>
>>
>> Hope my problem is better explained now.
>>
>> César
>>
>> James Bruck wrote:
>>> Hi César,
>>>
>>> The "exposing" of elements is handled by the TemplateParameter type.
>>> From the spec "...A TemplateParameter exposes ParameterableElements as
>>> formal parameters...".
>>> If I understand your question correctly, you want to specify the
>>> multiplicity of the property as a parameter. Unfortunately it appears
>>> that MultiplicityElement is not a ParameterableElement. That would seem
>>> to imply that one would not be able to specify
>>> TemplateParameterSubstitutions for some TemplateBinding to modify the
>>> bounds of the property.
>>> I took a quick look at all the RTF issues currently logged and this did
>>> not appear to be an issue that has been raised.
>>>
>>> The following might be of some use...
>>> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>>>
>>> I will dig more deeply into this tomorrow.
>>>
>>> Cheers,
>>> - James.
>>>
>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>> news:fsdbha$b4i$1@build.eclipse.org...
>>>> Hi,
>>>>
>>>> I have a problem with using templates and don't know if it is caused by
>>>> a real issue from the UML spec or if it is something I missed.
>>>>
>>>> A very frequent situation in our component-based modeling process cannot
>>>> be properly addressed by the UML templating mechanism as it is now. The
>>>> problem we face is similar to the one illustrated in the figure below
>>>> and concerns the fact that not always can the same parameter "expose"
>>>> (to use the UML lingo) two or more elements.
>>>>
>>>> ClassA<T, n>
>>>> prop1: T[n]
>>>> prop2: T[n]
>>>>
>>>> The example above works for parameter T, while it fails with n. Yet, the
>>>> need of having multiple elements whose multiplicity is parameterized by
>>>> a same parameter seems quite plausible. Taking a look at the UML
>>>> metamodel we can see that the references between a property and its
>>>> type, on one side, and between a property and its upper bound
>>>> multiplicity (i.e. upperValue), on the other side, present different
>>>> natures --the first case being a simple reference and the second, an
>>>> aggregation--, what makes all the difference when you want to expose the
>>>> same element multiple times. In our modeling process, most often than
>>>> not, a given parameter needs to be used by more than one element inside
>>>> a template.
>>>>
>>>> Can anyone comment on this?
>>>>
>>>> Thanks in advance
>>>>
>>>> César
>
|
|
| |
| Re: OMG issue tracking [message #477155 is a reply to message #477149] |
Fri, 28 March 2008 09:17 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Yes, you can. But this is a typical chicken-and-egg problem, isn't it? ;-)
Dave Carlson wrote:
> Last week at the Eclipse/OMG symposium, we had a brief discussion that the
> OMG should move to a more modern issue tracking system, like bugzilla. I
> find that the current OMG issue tracking is so archaic that I just don't
> bother using it. I was considering this same question earlier this morning
> as I uncovered several bugs in the OMG cmof metamodel for SBVR....
>
> Can we submit an issue to OMG to improve their issue tracking system :-)
>
> Dave
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:47EBC96A.1060008@yahoo.fr...
>>> You may want to raise an RTF issue with the OMG on this point.
>> Is there a reliable way of doing this? Last time I sent a mail to
>> issues@omg.org, they didn't even bother sending me an automatic reply, let
>> alone to file the issue.
>>
>
>
|
|
|
| Re: OMG issue tracking [message #477156 is a reply to message #477155] |
Fri, 28 March 2008 12:30 |
 Ed Merks
Messages: 33141
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------060801050901070503020804
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium.
http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...
kaiserlautern wrote:
> Yes, you can. But this is a typical chicken-and-egg problem, isn't it?
> ;-)
>
> Dave Carlson wrote:
>> Last week at the Eclipse/OMG symposium, we had a brief discussion
>> that the OMG should move to a more modern issue tracking system, like
>> bugzilla. I find that the current OMG issue tracking is so archaic
>> that I just don't bother using it. I was considering this same
>> question earlier this morning as I uncovered several bugs in the OMG
>> cmof metamodel for SBVR....
>>
>> Can we submit an issue to OMG to improve their issue tracking system :-)
>>
>> Dave
>>
>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>> news:47EBC96A.1060008@yahoo.fr...
>>>> You may want to raise an RTF issue with the OMG on this point.
>>> Is there a reliable way of doing this? Last time I sent a mail to
>>> issues@omg.org, they didn't even bother sending me an automatic
>>> reply, let alone to file the issue.
>>>
>>
>>
--------------060801050901070503020804
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium. <br>
<blockquote><a
href=" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l"> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l</a><br>
</blockquote>
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...<br>
<br>
<br>
kaiserlautern wrote:
<blockquote cite="mid:47ECB7B7.3000506@yahoo.fr" type="cite">Yes, you
can. But this is a typical chicken-and-egg problem, isn't it? ;-)
<br>
<br>
Dave Carlson wrote:
<br>
<blockquote type="cite">Last week at the Eclipse/OMG symposium, we
had a brief discussion that the OMG should move to a more modern issue
tracking system, like bugzilla. I find that the current OMG issue
tracking is so archaic that I just don't bother using it. I was
considering this same question earlier this morning as I uncovered
several bugs in the OMG cmof metamodel for SBVR....
<br>
<br>
Can we submit an issue to OMG to improve their issue tracking system
:-)
<br>
<br>
Dave
<br>
<br>
"kaiserlautern" <a class="moz-txt-link-rfc2396E" href="mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></a> wrote in message
<a class="moz-txt-link-freetext" href="news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</a>...
<br>
<blockquote type="cite">
<blockquote type="cite">You may want to raise an RTF issue with
the OMG on this point.
<br>
</blockquote>
Is there a reliable way of doing this? Last time I sent a mail to
<a class="moz-txt-link-abbreviated" href="mailto:issues@omg.org">issues@omg.org</a>, they didn't even bother sending me an automatic reply,
let alone to file the issue.
<br>
<br>
</blockquote>
<br>
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>
--------------060801050901070503020804--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
| |
| Re: OMG issue tracking [message #477188 is a reply to message #477156] |
Thu, 03 April 2008 21:17 |
Kenn Hussey
Messages: 1620
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
------=_NextPart_000_00C9_01C895AE.A72929C0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Ed,
Don't forget that we (as members of the OMG) ARE the OMG, in much the =
same way that we ARE Eclipse. The only chance we have for improvement is =
for individuals to step up and make things happen. I'm optimistic that =
the feedback we received at the first symposium (and should receive at =
the second one in June) may be the catalyst needed for change...
In the meantime, I don't believe that this particular issue needs to be =
raised to the OMG - see my response to James, above.
Kenn
"Ed Merks" <merks@ca.ibm.com> wrote in message =
news:fsioea$3d6$1@build.eclipse.org...
Issue tracking was one of the recurring themes at the EclipseCon OMG =
symposium. =20
http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
If issues aren't well tracked, they aren't likely to be well reported =
nor well resolved, so it's in the OMG's best interest to act...
kaiserlautern wrote:=20
Yes, you can. But this is a typical chicken-and-egg problem, isn't =
it? ;-)=20
Dave Carlson wrote:=20
Last week at the Eclipse/OMG symposium, we had a brief discussion =
that the OMG should move to a more modern issue tracking system, like =
bugzilla. I find that the current OMG issue tracking is so archaic that =
I just don't bother using it. I was considering this same question =
earlier this morning as I uncovered several bugs in the OMG cmof =
metamodel for SBVR....=20
Can we submit an issue to OMG to improve their issue tracking =
system :-)=20
Dave=20
"kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message =
news:47EBC96A.1060008@yahoo.fr...=20
You may want to raise an RTF issue with the OMG on this point. =
Is there a reliable way of doing this? Last time I sent a mail =
to issues@omg.org, they didn't even bother sending me an automatic =
reply, let alone to file the issue.=20
------=_NextPart_000_00C9_01C895AE.A72929C0
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type =
content=3Dtext/html;charset=3DISO-8859-1>
<META content=3D"MSHTML 6.00.6000.16608" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY text=3D#000000 bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2>Ed,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Don't forget that we (as members =
of the OMG)=20
ARE the OMG, in much the same way that we ARE Eclipse. The only =
chance we=20
have for improvement is for individuals to step up and make =
things=20
happen. I'm optimistic that the feedback we received at the first =
symposium=20
(and should receive at the second one in June) may be the catalyst =
needed for=20
change...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>In the meantime, I don't believe that =
this=20
particular issue needs to be raised to the OMG - see my =
response to=20
James, above.</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Kenn</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
<DIV>"Ed Merks" <<A =
href=3D"mailto:merks@ca.ibm.com">merks@ca.ibm.com</A>>=20
wrote in message <A=20
=
href=3D"news:fsioea$3d6$1@build.eclipse.org">news:fsioea$3d6$1@build.ecli=
pse.org</A>...</DIV>Issue=20
tracking was one of the recurring themes at the EclipseCon OMG=20
symposium. <BR>
<BLOCKQUOTE><A=20
=
href=3D" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l">ht=
tp://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.html</A><BR></BLOC=
KQUOTE>If=20
issues aren't well tracked, they aren't likely to be well reported nor =
well=20
resolved, so it's in the OMG's best interest to=20
act...<BR><BR><BR>kaiserlautern wrote:=20
<BLOCKQUOTE cite=3Dmid:47ECB7B7.3000506@yahoo.fr type=3D"cite">Yes, =
you can. But=20
this is a typical chicken-and-egg problem, isn't it? ;-) =
<BR><BR>Dave=20
Carlson wrote: <BR>
<BLOCKQUOTE type=3D"cite">Last week at the Eclipse/OMG symposium, we =
had a=20
brief discussion that the OMG should move to a more modern issue =
tracking=20
system, like bugzilla. I find that the current OMG issue =
tracking is=20
so archaic that I just don't bother using it. I was =
considering this=20
same question earlier this morning as I uncovered several bugs in =
the OMG=20
cmof metamodel for SBVR.... <BR><BR>Can we submit an issue to OMG =
to=20
improve their issue tracking system :-) <BR><BR>Dave=20
<BR><BR>"kaiserlautern" <A class=3Dmoz-txt-link-rfc2396E=20
=
href=3D"mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></A>=
=20
wrote in message <A class=3Dmoz-txt-link-freetext=20
=
href=3D"news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</A=
>...=20
<BR>
<BLOCKQUOTE type=3D"cite">
<BLOCKQUOTE type=3D"cite">You may want to raise an RTF issue =
with the=20
OMG on this point. <BR></BLOCKQUOTE>Is there a reliable way of =
doing=20
this? Last time I sent a mail to <A =
class=3Dmoz-txt-link-abbreviated=20
href=3D"mailto:issues@omg.org">issues@omg.org</A>, they didn't =
even bother=20
sending me an automatic reply, let alone to file the issue.=20
=
<BR><BR></BLOCKQUOTE><BR><BR></BLOCKQUOTE></BLOCKQUOTE ><BR></BLOCKQUOTE><=
/BODY></HTML>
------=_NextPart_000_00C9_01C895AE.A72929C0--
|
|
|
| Re: OMG issue tracking [message #477189 is a reply to message #477188] |
Fri, 04 April 2008 11:19 |
Ed Merks
Messages: 33141
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------070002040706030708080501
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Kenn,
Really, I'm a member? And what exactly do you mean by implying that
whining isn't sufficient? It's so much easier than having to take
responsibility... :-P
Kenn Hussey wrote:
> Ed,
>
> Don't forget that we (as members of the OMG) ARE the OMG, in much the
> same way that we ARE Eclipse. The only chance we have for
> improvement is for individuals to step up and make things happen. I'm
> optimistic that the feedback we received at the first symposium (and
> should receive at the second one in June) may be the catalyst needed
> for change...
>
> In the meantime, I don't believe that this particular issue needs to
> be raised to the OMG - see my response to James, above.
>
> Kenn
>
>
> "Ed Merks" <merks@ca.ibm.com <mailto:merks@ca.ibm.com>> wrote in
> message news:fsioea$3d6$1@build.eclipse.org...
> Issue tracking was one of the recurring themes at the EclipseCon
> OMG symposium.
>
> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
>
> If issues aren't well tracked, they aren't likely to be well
> reported nor well resolved, so it's in the OMG's best interest to
> act...
>
>
> kaiserlautern wrote:
>> Yes, you can. But this is a typical chicken-and-egg problem,
>> isn't it? ;-)
>>
>> Dave Carlson wrote:
>>> Last week at the Eclipse/OMG symposium, we had a brief
>>> discussion that the OMG should move to a more modern issue
>>> tracking system, like bugzilla. I find that the current OMG
>>> issue tracking is so archaic that I just don't bother using it.
>>> I was considering this same question earlier this morning as I
>>> uncovered several bugs in the OMG cmof metamodel for SBVR....
>>>
>>> Can we submit an issue to OMG to improve their issue tracking
>>> system :-)
>>>
>>> Dave
>>>
>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>> news:47EBC96A.1060008@yahoo.fr...
>>>>> You may want to raise an RTF issue with the OMG on this point.
>>>> Is there a reliable way of doing this? Last time I sent a mail
>>>> to issues@omg.org, they didn't even bother sending me an
>>>> automatic reply, let alone to file the issue.
>>>>
>>>
>>>
>
--------------070002040706030708080501
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Kenn,<br>
<br>
Really, I'm a member? And what exactly do you mean by implying that
whining isn't sufficient? It's so much easier than having to take
responsibility... :-P<br>
<br>
<br>
Kenn Hussey wrote:
<blockquote cite="mid:ft466a$bbn$3@build.eclipse.org" type="cite">
<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">
<meta content="MSHTML 6.00.6000.16608" name="GENERATOR">
<style></style>
<div><font face="Arial" size="2">Ed,</font></div>
<div> </div>
<div><font face="Arial" size="2">Don't forget that we (as members of
the OMG) ARE the OMG, in much the same way that we ARE Eclipse. The
only chance we have for improvement is for individuals to step up and
make things happen. I'm optimistic that the feedback we received at the
first symposium (and should receive at the second one in June) may be
the catalyst needed for change...</font></div>
<div> </div>
<div><font face="Arial" size="2">In the meantime, I don't believe
that this particular issue needs to be raised to the OMG - see my
response to James, above.</font></div>
<div> </div>
<div><font face="Arial" size="2">Kenn</font></div>
<div> </div>
<blockquote
style="border-left: 2px solid rgb(0, 0, 0); padding-right: 0px; padding-left: 5px; margin-left: 5px; margin-right: 0px;">
<div>"Ed Merks" <<a moz-do-not-send="true"
href="mailto:merks@ca.ibm.com">merks@ca.ibm.com</a>> wrote in
message <a moz-do-not-send="true"
href="news:fsioea$3d6$1@build.eclipse.org">news:fsioea$3d6$1@build.eclipse.org</a>...</div>
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium. <br>
<blockquote><a moz-do-not-send="true"
href=" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l"> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l</a><br>
</blockquote>
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...<br>
<br>
<br>
kaiserlautern wrote:
<blockquote cite="mid:47ECB7B7.3000506@yahoo.fr" type="cite">Yes,
you can. But this is a typical chicken-and-egg problem, isn't it? ;-) <br>
<br>
Dave Carlson wrote: <br>
<blockquote type="cite">Last week at the Eclipse/OMG symposium,
we had a brief discussion that the OMG should move to a more modern
issue tracking system, like bugzilla. I find that the current OMG
issue tracking is so archaic that I just don't bother using it. I was
considering this same question earlier this morning as I uncovered
several bugs in the OMG cmof metamodel for SBVR.... <br>
<br>
Can we submit an issue to OMG to improve their issue tracking system
:-) <br>
<br>
Dave <br>
<br>
"kaiserlautern" <a moz-do-not-send="true" class="moz-txt-link-rfc2396E"
href="mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></a>
wrote in message <a moz-do-not-send="true"
class="moz-txt-link-freetext" href="news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</a>...
<br>
<blockquote type="cite">
<blockquote type="cite">You may want to raise an RTF issue
with the OMG on this point. <br>
</blockquote>
Is there a reliable way of doing this? Last time I sent a mail to <a
moz-do-not-send="true" class="moz-txt-link-abbreviated"
href="mailto:issues@omg.org">issues@omg.org</a>, they didn't even
bother sending me an automatic reply, let alone to file the issue. <br>
<br>
</blockquote>
<br>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>
--------------070002040706030708080501--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
| | |
| Re: UML templates [message #477202 is a reply to message #477200] |
Fri, 04 April 2008 18:57 |
james bruck
Messages: 1724
Registered: July 2009 |
Senior Member |
|
|
Hi Kenn,
Say I had two Properties in a TemplateableElement, n and m. I want to say
that the type of n is "E" and the type of m is "E". So we have n:E and
m:E. E is a parameterable element of type Class owned by some template
parameter. When I then bind formal parameter "E" to some actual parameter
"Car" I would then have n:Car and m:Car (in the bound class).
In this example, we have one template parameter used in several places in
the template which is great, however, I wouldn't be able to do the same
thing with multiplicities since the ValueSpecification representing upper
and lower values would have to be owned by only one element. I think this
is a bit awkward.
... or am I missing something?
Cheers,
- James.
"Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
news:ft5guo$c0c$1@build.eclipse.org...
> James,
>
> You are correct, this it how it works - each entity (whether it be a type
> or a value specification) must be exposed using a separate template
> parameter. I don't personally see an issue with this...
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:ft5edu$n98$3@build.eclipse.org...
>> Hi Kenn,
>>
>> It is true that the TemplateParameter does not have to own its
>> ParameterableElements so the TemplateParameter could reference the
>> multiplicity value specification for a property, however, the problem
>> arises when we have two or more properties whose multiplicity should
>> change as a result of TemplateParameterSubstitution.
>> Since the ValueSpecification (ParameterableElement) representing the
>> multiplicity is directly owned by only one entity, we could not express
>> the fact that we want to substitute the multiplicities of many properties
>> with one template parameter.
>>
>> Contrast this with changing the Type of a property via
>> TemplateParameterSubstitution. A property references its Type (does not
>> directly own it) so it does not suffer from this issue. We can for
>> example have the TemplateParameter own the ParameterableElement
>> representing the type and then formulate multiple substitutions...no
>> problem.
>>
>>
>> ... Please correct me if I'm wrong.
>>
>> Cheers,
>>
>> - James.
>>
>>
>>
>>
>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>> news:ft4669$bbn$1@build.eclipse.org...
>>> Wait a minute. A template parameter need not own its parameterable
>>> elements... why don't you just reference the value specification via
>>> TemplateParameter::parameteredElement?
>>>
>>> Kenn
>>>
>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>> Hi C
|
|
|
| Re: UML templates [message #477203 is a reply to message #477202] |
Mon, 07 April 2008 13:38 |
Kenn Hussey
Messages: 1620
Registered: July 2009 |
Senior Member |
|
|
James,
Yes, it's a bit awkward, but that's the difference between copy by reference
and copy by value...
Kenn
"James Bruck" <jbruck@ca.ibm.com> wrote in message
news:ft5tmo$sgk$3@build.eclipse.org...
> Hi Kenn,
>
> Say I had two Properties in a TemplateableElement, n and m. I want to
> say that the type of n is "E" and the type of m is "E". So we have n:E
> and m:E. E is a parameterable element of type Class owned by some
> template parameter. When I then bind formal parameter "E" to some
> actual parameter "Car" I would then have n:Car and m:Car (in the bound
> class).
> In this example, we have one template parameter used in several places in
> the template which is great, however, I wouldn't be able to do the same
> thing with multiplicities since the ValueSpecification representing upper
> and lower values would have to be owned by only one element. I think
> this is a bit awkward.
>
> .. or am I missing something?
>
> Cheers,
> - James.
>
>
>
> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
> news:ft5guo$c0c$1@build.eclipse.org...
>> James,
>>
>> You are correct, this it how it works - each entity (whether it be a type
>> or a value specification) must be exposed using a separate template
>> parameter. I don't personally see an issue with this...
>>
>> Kenn
>>
>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>> news:ft5edu$n98$3@build.eclipse.org...
>>> Hi Kenn,
>>>
>>> It is true that the TemplateParameter does not have to own its
>>> ParameterableElements so the TemplateParameter could reference the
>>> multiplicity value specification for a property, however, the problem
>>> arises when we have two or more properties whose multiplicity should
>>> change as a result of TemplateParameterSubstitution.
>>> Since the ValueSpecification (ParameterableElement) representing the
>>> multiplicity is directly owned by only one entity, we could not express
>>> the fact that we want to substitute the multiplicities of many
>>> properties with one template parameter.
>>>
>>> Contrast this with changing the Type of a property via
>>> TemplateParameterSubstitution. A property references its Type (does not
>>> directly own it) so it does not suffer from this issue. We can for
>>> example have the TemplateParameter own the ParameterableElement
>>> representing the type and then formulate multiple substitutions...no
>>> problem.
>>>
>>>
>>> ... Please correct me if I'm wrong.
>>>
>>> Cheers,
>>>
>>> - James.
>>>
>>>
>>>
>>>
>>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>>> news:ft4669$bbn$1@build.eclipse.org...
>>>> Wait a minute. A template parameter need not own its parameterable
>>>> elements... why don't you just reference the value specification via
>>>> TemplateParameter::parameteredElement?
>>>>
>>>> Kenn
>>>>
>>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>>> Hi C
|
|
|
| Re: UML templates [message #477282 is a reply to message #477203] |
Thu, 24 April 2008 20:06 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Sorry for the delay. I've been busy for a while with the m2m (QVTO) forum
;-). In fact, our need in terms of parameterizations is very
straightforward: we usually have components with sometimes dozens of
multiplicityElements inside (ports, parts, connectorEnds) and we cannot
afford to hang dozens of parameters to the same signature when - usually -
all multiplicities should change synchronously. In this case, a single
parameter is, most of the time, enough. As a workaround, I created a
NamedParameter stereotype so that internal elements can refer to parameters
by their names, but ideally, it would be better if MultiplicityElement had
also simple references to ValueSpecification (in addition to the
aggregation-type lowerValue and upperValue). By the way, did anybody find a
good reason for a TemplateParameter not to extend NamedElement? ;-)
César
"Kenn Hussey" <Kenn.Hussey@embarcadero.com> a écrit dans le message de
news:ftd84j$35a$1@build.eclipse.org...
> James,
>
> Yes, it's a bit awkward, but that's the difference between copy by
> reference and copy by value...
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:ft5tmo$sgk$3@build.eclipse.org...
>> Hi Kenn,
>>
>> Say I had two Properties in a TemplateableElement, n and m. I want to
>> say that the type of n is "E" and the type of m is "E". So we have n:E
>> and m:E. E is a parameterable element of type Class owned by some
>> template parameter. When I then bind formal parameter "E" to some
>> actual parameter "Car" I would then have n:Car and m:Car (in the bound
>> class).
>> In this example, we have one template parameter used in several places in
>> the template which is great, however, I wouldn't be able to do the same
>> thing with multiplicities since the ValueSpecification representing upper
>> and lower values would have to be owned by only one element. I think
>> this is a bit awkward.
>>
>> .. or am I missing something?
>>
>> Cheers,
>> - James.
>>
>>
>>
>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>> news:ft5guo$c0c$1@build.eclipse.org...
>>> James,
>>>
>>> You are correct, this it how it works - each entity (whether it be a
>>> type or a value specification) must be exposed using a separate template
>>> parameter. I don't personally see an issue with this...
>>>
>>> Kenn
>>>
>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>> news:ft5edu$n98$3@build.eclipse.org...
>>>> Hi Kenn,
>>>>
>>>> It is true that the TemplateParameter does not have to own its
>>>> ParameterableElements so the TemplateParameter could reference the
>>>> multiplicity value specification for a property, however, the problem
>>>> arises when we have two or more properties whose multiplicity should
>>>> change as a result of TemplateParameterSubstitution.
>>>> Since the ValueSpecification (ParameterableElement) representing the
>>>> multiplicity is directly owned by only one entity, we could not express
>>>> the fact that we want to substitute the multiplicities of many
>>>> properties with one template parameter.
>>>>
>>>> Contrast this with changing the Type of a property via
>>>> TemplateParameterSubstitution. A property references its Type (does
>>>> not directly own it) so it does not suffer from this issue. We can for
>>>> example have the TemplateParameter own the ParameterableElement
>>>> representing the type and then formulate multiple substitutions...no
>>>> problem.
>>>>
>>>>
>>>> ... Please correct me if I'm wrong.
>>>>
>>>> Cheers,
>>>>
>>>> - James.
>>>>
>>>>
>>>>
>>>>
>>>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>>>> news:ft4669$bbn$1@build.eclipse.org...
>>>>> Wait a minute. A template parameter need not own its parameterable
>>>>> elements... why don't you just reference the value specification via
>>>>> TemplateParameter::parameteredElement?
>>>>>
>>>>> Kenn
>>>>>
>>>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>>>> Hi César,
>>>>>>
>>>>>> The second posting clarified it for me. I see the issue.
>>>>>> Unfortunately it appears that you will not be able to reuse your
>>>>>> value specification.
>>>>>> As you mentioned, an element cannot be owned by more than one
>>>>>> container. This seems to be a basic problem that UML should address
>>>>>> or at the very least it should be brought to the attention of those
>>>>>> defining the spec.
>>>>>>
>>>>>> You may want to raise an RTF issue with the OMG on this point.
>>>>>>
>>>>>> Cheers,
>>>>>> - James.
>>>>>>
>>>>>>
>>>>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>>>>> news:47EB85BA.4050409@yahoo.fr...
>>>>>>> Hi, James,
>>>>>>>
>>>>>>> Thanks for your answer. By the way, I had already read your good
>>>>>>> article on generics.
>>>>>>>
>>>>>>> I had also looked for related questions in the UML issues, but to no
>>>>>>> avail. In fact what I need to "expose" is not the property
>>>>>>> (MultiplicityElement) but its multiplicity (i.e. a
>>>>>>> ValueSpecification). And according to the UML spec, value
>>>>>>> specifications can be exposed. As I said before, a parameter can
>>>>>>> expose only one multiplicity value while I needed to have the same
>>>>>>> parameter exposing more than one multiplicity value.
>>>>>>>
>>>>>>> To solve this problem, I use the aggregation type
>>>>>>> ownedParametedElement (i.e. the parameterable element belongs to the
>>>>>>> template parameter). And this parameterable element, in this case,
>>>>>>> is my value specification (say, vs1) that will be used as a value
>>>>>>> for my exposed multiplicities. However, when I specify that my
>>>>>>> property1 has a multiplicity value (upperValue) equals to vs1, it
>>>>>>> takes possession of it and I cannot reuse vs1 as an upperValue for
>>>>>>> another property (because an element cannot be owned by more than
>>>>>>> one "container"). And this contrasts with the relationship between a
>>>>>>> property and its type, which is just a simple reference (that's why
>>>>>>> in my simple example we can have more than one property with the
>>>>>>> same type T)
>>>>>>>
>>>>>>> Schematically, we would have (please make an extra effort to
>>>>>>> understand these diagrams ;-)
>>>>>>>
>>>>>>> ------------------------
>>>>>>> ParameterElement
>>>>>>> ------------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ParameterableElement
>>>>>>> ----------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ValueSpecification
>>>>>>> ----------------------
>>>>>>> ^
>>>>>>> | (aggregation)
>>>>>>> <>
>>>>>>> ------------------------
>>>>>>> MultiplicityElement
>>>>>>> -----------------------
>>>>>>>
>>>>>>> ---------------------------------------------------------
>>>>>>>
>>>>>>> ------------------------
>>>>>>> ParameterElement
>>>>>>> ------------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ParameterableElement
>>>>>>> ----------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> Classifier
>>>>>>> ----------------------
>>>>>>> ^ ^
>>>>>>> | | (simple references)
>>>>>>> | |
>>>>>>> --------------- --------------
>>>>>>> Property1 Property2
>>>>>>> --------------- --------------
>>>>>>>
>>>>>>>
>>>>>>> Hope my problem is better explained now.
>>>>>>>
>>>>>>> César
>>>>>>>
>>>>>>> James Bruck wrote:
>>>>>>>> Hi César,
>>>>>>>>
>>>>>>>> The "exposing" of elements is handled by the TemplateParameter
>>>>>>>> type. From the spec "...A TemplateParameter exposes
>>>>>>>> ParameterableElements as formal parameters...".
>>>>>>>> If I understand your question correctly, you want to specify the
>>>>>>>> multiplicity of the property as a parameter. Unfortunately it
>>>>>>>> appears that MultiplicityElement is not a ParameterableElement.
>>>>>>>> That would seem to imply that one would not be able to specify
>>>>>>>> TemplateParameterSubstitutions for some TemplateBinding to modify
>>>>>>>> the bounds of the property.
>>>>>>>> I took a quick look at all the RTF issues currently logged and this
>>>>>>>> did not appear to be an issue that has been raised.
>>>>>>>>
>>>>>>>> The following might be of some use...
>>>>>>>> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>>>>>>>>
>>>>>>>> I will dig more deeply into this tomorrow.
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> - James.
>>>>>>>>
>>>>>>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>>>>>>> news:fsdbha$b4i$1@build.eclipse.org...
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> I have a problem with using templates and don't know if it is
>>>>>>>>> caused by a real issue from the UML spec or if it is something I
>>>>>>>>> missed.
>>>>>>>>>
>>>>>>>>> A very frequent situation in our component-based modeling process
>>>>>>>>> cannot be properly addressed by the UML templating mechanism as it
>>>>>>>>> is now. The problem we face is similar to the one illustrated in
>>>>>>>>> the figure below and concerns the fact that not always can the
>>>>>>>>> same parameter "expose" (to use the UML lingo) two or more
>>>>>>>>> elements.
>>>>>>>>>
>>>>>>>>> ClassA<T, n>
>>>>>>>>> prop1: T[n]
>>>>>>>>> prop2: T[n]
>>>>>>>>>
>>>>>>>>> The example above works for parameter T, while it fails with n.
>>>>>>>>> Yet, the need of having multiple elements whose multiplicity is
>>>>>>>>> parameterized by a same parameter seems quite plausible. Taking a
>>>>>>>>> look at the UML metamodel we can see that the references between a
>>>>>>>>> property and its type, on one side, and between a property and its
>>>>>>>>> upper bound multiplicity (i.e. upperValue), on the other side,
>>>>>>>>> present different natures --the first case being a simple
>>>>>>>>> reference and the second, an aggregation--, what makes all the
>>>>>>>>> difference when you want to expose the same element multiple
>>>>>>>>> times. In our modeling process, most often than not, a given
>>>>>>>>> parameter needs to be used by more than one element inside a
>>>>>>>>> template.
>>>>>>>>>
>>>>>>>>> Can anyone comment on this?
>>>>>>>>>
>>>>>>>>> Thanks in advance
>>>>>>>>>
>>>>>>>>> César
>>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>
|
|
| | |
| Re: UML templates [message #626275 is a reply to message #477139] |
Thu, 27 March 2008 11:32 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Hi, James,
Thanks for your answer. By the way, I had already read your good article
on generics.
I had also looked for related questions in the UML issues, but to no
avail. In fact what I need to "expose" is not the property
(MultiplicityElement) but its multiplicity (i.e. a ValueSpecification).
And according to the UML spec, value specifications can be exposed. As I
said before, a parameter can expose only one multiplicity value while I
needed to have the same parameter exposing more than one multiplicity
value.
To solve this problem, I use the aggregation type ownedParametedElement
(i.e. the parameterable element belongs to the template parameter). And
this parameterable element, in this case, is my value specification
(say, vs1) that will be used as a value for my exposed multiplicities.
However, when I specify that my property1 has a multiplicity value
(upperValue) equals to vs1, it takes possession of it and I cannot reuse
vs1 as an upperValue for another property (because an element cannot be
owned by more than one "container"). And this contrasts with the
relationship between a property and its type, which is just a simple
reference (that's why in my simple example we can have more than one
property with the same type T)
Schematically, we would have (please make an extra effort to understand
these diagrams ;-)
------------------------
ParameterElement
------------------------
<>
|
v
-----------------------
ParameterableElement
----------------------
<>
|
v
-----------------------
ValueSpecification
----------------------
^
| (aggregation)
<>
------------------------
MultiplicityElement
-----------------------
---------------------------------------------------------
------------------------
ParameterElement
------------------------
<>
|
v
-----------------------
ParameterableElement
----------------------
<>
|
v
-----------------------
Classifier
----------------------
^ ^
| | (simple references)
| |
--------------- --------------
Property1 Property2
--------------- --------------
Hope my problem is better explained now.
César
James Bruck wrote:
> Hi César,
>
> The "exposing" of elements is handled by the TemplateParameter type. From
> the spec "...A TemplateParameter exposes ParameterableElements as formal
> parameters...".
> If I understand your question correctly, you want to specify the
> multiplicity of the property as a parameter. Unfortunately it appears that
> MultiplicityElement is not a ParameterableElement. That would seem to
> imply that one would not be able to specify TemplateParameterSubstitutions
> for some TemplateBinding to modify the bounds of the property.
> I took a quick look at all the RTF issues currently logged and this did not
> appear to be an issue that has been raised.
>
> The following might be of some use...
> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>
> I will dig more deeply into this tomorrow.
>
> Cheers,
> - James.
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:fsdbha$b4i$1@build.eclipse.org...
>> Hi,
>>
>> I have a problem with using templates and don't know if it is caused by a
>> real issue from the UML spec or if it is something I missed.
>>
>> A very frequent situation in our component-based modeling process cannot
>> be properly addressed by the UML templating mechanism as it is now. The
>> problem we face is similar to the one illustrated in the figure below and
>> concerns the fact that not always can the same parameter "expose" (to use
>> the UML lingo) two or more elements.
>>
>> ClassA<T, n>
>> prop1: T[n]
>> prop2: T[n]
>>
>> The example above works for parameter T, while it fails with n. Yet, the
>> need of having multiple elements whose multiplicity is parameterized by a
>> same parameter seems quite plausible. Taking a look at the UML metamodel
>> we can see that the references between a property and its type, on one
>> side, and between a property and its upper bound multiplicity (i.e.
>> upperValue), on the other side, present different natures --the first case
>> being a simple reference and the second, an aggregation--, what makes all
>> the difference when you want to expose the same element multiple times. In
>> our modeling process, most often than not, a given parameter needs to be
>> used by more than one element inside a template.
>>
>> Can anyone comment on this?
>>
>> Thanks in advance
>>
>> César
>
>
|
|
| |
| Re: UML templates [message #626282 is a reply to message #477142] |
Thu, 27 March 2008 16:20 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
> You may want to raise an RTF issue with the OMG on this point.
Is there a reliable way of doing this? Last time I sent a mail to
issues@omg.org, they didn't even bother sending me an automatic reply,
let alone to file the issue.
Thanks for your help anyway.
César
James Bruck wrote:
> Hi César,
>
> The second posting clarified it for me. I see the issue. Unfortunately it
> appears that you will not be able to reuse your value specification.
> As you mentioned, an element cannot be owned by more than one container.
> This seems to be a basic problem that UML should address or at the very
> least it should be brought to the attention of those defining the spec.
>
> You may want to raise an RTF issue with the OMG on this point.
>
> Cheers,
> - James.
>
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:47EB85BA.4050409@yahoo.fr...
>> Hi, James,
>>
>> Thanks for your answer. By the way, I had already read your good article
>> on generics.
>>
>> I had also looked for related questions in the UML issues, but to no
>> avail. In fact what I need to "expose" is not the property
>> (MultiplicityElement) but its multiplicity (i.e. a ValueSpecification).
>> And according to the UML spec, value specifications can be exposed. As I
>> said before, a parameter can expose only one multiplicity value while I
>> needed to have the same parameter exposing more than one multiplicity
>> value.
>>
>> To solve this problem, I use the aggregation type ownedParametedElement
>> (i.e. the parameterable element belongs to the template parameter). And
>> this parameterable element, in this case, is my value specification (say,
>> vs1) that will be used as a value for my exposed multiplicities. However,
>> when I specify that my property1 has a multiplicity value (upperValue)
>> equals to vs1, it takes possession of it and I cannot reuse vs1 as an
>> upperValue for another property (because an element cannot be owned by
>> more than one "container"). And this contrasts with the relationship
>> between a property and its type, which is just a simple reference (that's
>> why in my simple example we can have more than one property with the same
>> type T)
>>
>> Schematically, we would have (please make an extra effort to understand
>> these diagrams ;-)
>>
>> ------------------------
>> ParameterElement
>> ------------------------
>> <>
>> |
>> v
>> -----------------------
>> ParameterableElement
>> ----------------------
>> <>
>> |
>> v
>> -----------------------
>> ValueSpecification
>> ----------------------
>> ^
>> | (aggregation)
>> <>
>> ------------------------
>> MultiplicityElement
>> -----------------------
>>
>> ---------------------------------------------------------
>>
>> ------------------------
>> ParameterElement
>> ------------------------
>> <>
>> |
>> v
>> -----------------------
>> ParameterableElement
>> ----------------------
>> <>
>> |
>> v
>> -----------------------
>> Classifier
>> ----------------------
>> ^ ^
>> | | (simple references)
>> | |
>> --------------- --------------
>> Property1 Property2
>> --------------- --------------
>>
>>
>> Hope my problem is better explained now.
>>
>> César
>>
>> James Bruck wrote:
>>> Hi César,
>>>
>>> The "exposing" of elements is handled by the TemplateParameter type.
>>> From the spec "...A TemplateParameter exposes ParameterableElements as
>>> formal parameters...".
>>> If I understand your question correctly, you want to specify the
>>> multiplicity of the property as a parameter. Unfortunately it appears
>>> that MultiplicityElement is not a ParameterableElement. That would seem
>>> to imply that one would not be able to specify
>>> TemplateParameterSubstitutions for some TemplateBinding to modify the
>>> bounds of the property.
>>> I took a quick look at all the RTF issues currently logged and this did
>>> not appear to be an issue that has been raised.
>>>
>>> The following might be of some use...
>>> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>>>
>>> I will dig more deeply into this tomorrow.
>>>
>>> Cheers,
>>> - James.
>>>
>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>> news:fsdbha$b4i$1@build.eclipse.org...
>>>> Hi,
>>>>
>>>> I have a problem with using templates and don't know if it is caused by
>>>> a real issue from the UML spec or if it is something I missed.
>>>>
>>>> A very frequent situation in our component-based modeling process cannot
>>>> be properly addressed by the UML templating mechanism as it is now. The
>>>> problem we face is similar to the one illustrated in the figure below
>>>> and concerns the fact that not always can the same parameter "expose"
>>>> (to use the UML lingo) two or more elements.
>>>>
>>>> ClassA<T, n>
>>>> prop1: T[n]
>>>> prop2: T[n]
>>>>
>>>> The example above works for parameter T, while it fails with n. Yet, the
>>>> need of having multiple elements whose multiplicity is parameterized by
>>>> a same parameter seems quite plausible. Taking a look at the UML
>>>> metamodel we can see that the references between a property and its
>>>> type, on one side, and between a property and its upper bound
>>>> multiplicity (i.e. upperValue), on the other side, present different
>>>> natures --the first case being a simple reference and the second, an
>>>> aggregation--, what makes all the difference when you want to expose the
>>>> same element multiple times. In our modeling process, most often than
>>>> not, a given parameter needs to be used by more than one element inside
>>>> a template.
>>>>
>>>> Can anyone comment on this?
>>>>
>>>> Thanks in advance
>>>>
>>>> César
>
|
|
| |
| Re: OMG issue tracking [message #626297 is a reply to message #477149] |
Fri, 28 March 2008 09:17 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Yes, you can. But this is a typical chicken-and-egg problem, isn't it? ;-)
Dave Carlson wrote:
> Last week at the Eclipse/OMG symposium, we had a brief discussion that the
> OMG should move to a more modern issue tracking system, like bugzilla. I
> find that the current OMG issue tracking is so archaic that I just don't
> bother using it. I was considering this same question earlier this morning
> as I uncovered several bugs in the OMG cmof metamodel for SBVR....
>
> Can we submit an issue to OMG to improve their issue tracking system :-)
>
> Dave
>
> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
> news:47EBC96A.1060008@yahoo.fr...
>>> You may want to raise an RTF issue with the OMG on this point.
>> Is there a reliable way of doing this? Last time I sent a mail to
>> issues@omg.org, they didn't even bother sending me an automatic reply, let
>> alone to file the issue.
>>
>
>
|
|
|
| Re: OMG issue tracking [message #626298 is a reply to message #477155] |
Fri, 28 March 2008 12:30 |
Ed Merks
Messages: 33141
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------060801050901070503020804
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium.
http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...
kaiserlautern wrote:
> Yes, you can. But this is a typical chicken-and-egg problem, isn't it?
> ;-)
>
> Dave Carlson wrote:
>> Last week at the Eclipse/OMG symposium, we had a brief discussion
>> that the OMG should move to a more modern issue tracking system, like
>> bugzilla. I find that the current OMG issue tracking is so archaic
>> that I just don't bother using it. I was considering this same
>> question earlier this morning as I uncovered several bugs in the OMG
>> cmof metamodel for SBVR....
>>
>> Can we submit an issue to OMG to improve their issue tracking system :-)
>>
>> Dave
>>
>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>> news:47EBC96A.1060008@yahoo.fr...
>>>> You may want to raise an RTF issue with the OMG on this point.
>>> Is there a reliable way of doing this? Last time I sent a mail to
>>> issues@omg.org, they didn't even bother sending me an automatic
>>> reply, let alone to file the issue.
>>>
>>
>>
--------------060801050901070503020804
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium. <br>
<blockquote><a
href=" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l"> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l</a><br>
</blockquote>
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...<br>
<br>
<br>
kaiserlautern wrote:
<blockquote cite="mid:47ECB7B7.3000506@yahoo.fr" type="cite">Yes, you
can. But this is a typical chicken-and-egg problem, isn't it? ;-)
<br>
<br>
Dave Carlson wrote:
<br>
<blockquote type="cite">Last week at the Eclipse/OMG symposium, we
had a brief discussion that the OMG should move to a more modern issue
tracking system, like bugzilla. I find that the current OMG issue
tracking is so archaic that I just don't bother using it. I was
considering this same question earlier this morning as I uncovered
several bugs in the OMG cmof metamodel for SBVR....
<br>
<br>
Can we submit an issue to OMG to improve their issue tracking system
:-)
<br>
<br>
Dave
<br>
<br>
"kaiserlautern" <a class="moz-txt-link-rfc2396E" href="mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></a> wrote in message
<a class="moz-txt-link-freetext" href="news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</a>...
<br>
<blockquote type="cite">
<blockquote type="cite">You may want to raise an RTF issue with
the OMG on this point.
<br>
</blockquote>
Is there a reliable way of doing this? Last time I sent a mail to
<a class="moz-txt-link-abbreviated" href="mailto:issues@omg.org">issues@omg.org</a>, they didn't even bother sending me an automatic reply,
let alone to file the issue.
<br>
<br>
</blockquote>
<br>
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>
--------------060801050901070503020804--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
| |
| Re: OMG issue tracking [message #626348 is a reply to message #477156] |
Thu, 03 April 2008 21:17 |
Kenn Hussey
Messages: 1620
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
------=_NextPart_000_00C9_01C895AE.A72929C0
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Ed,
Don't forget that we (as members of the OMG) ARE the OMG, in much the =
same way that we ARE Eclipse. The only chance we have for improvement is =
for individuals to step up and make things happen. I'm optimistic that =
the feedback we received at the first symposium (and should receive at =
the second one in June) may be the catalyst needed for change...
In the meantime, I don't believe that this particular issue needs to be =
raised to the OMG - see my response to James, above.
Kenn
"Ed Merks" <merks@ca.ibm.com> wrote in message =
news:fsioea$3d6$1@build.eclipse.org...
Issue tracking was one of the recurring themes at the EclipseCon OMG =
symposium. =20
http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
If issues aren't well tracked, they aren't likely to be well reported =
nor well resolved, so it's in the OMG's best interest to act...
kaiserlautern wrote:=20
Yes, you can. But this is a typical chicken-and-egg problem, isn't =
it? ;-)=20
Dave Carlson wrote:=20
Last week at the Eclipse/OMG symposium, we had a brief discussion =
that the OMG should move to a more modern issue tracking system, like =
bugzilla. I find that the current OMG issue tracking is so archaic that =
I just don't bother using it. I was considering this same question =
earlier this morning as I uncovered several bugs in the OMG cmof =
metamodel for SBVR....=20
Can we submit an issue to OMG to improve their issue tracking =
system :-)=20
Dave=20
"kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message =
news:47EBC96A.1060008@yahoo.fr...=20
You may want to raise an RTF issue with the OMG on this point. =
Is there a reliable way of doing this? Last time I sent a mail =
to issues@omg.org, they didn't even bother sending me an automatic =
reply, let alone to file the issue.=20
------=_NextPart_000_00C9_01C895AE.A72929C0
Content-Type: text/html;
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type =
content=3Dtext/html;charset=3DISO-8859-1>
<META content=3D"MSHTML 6.00.6000.16608" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY text=3D#000000 bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2>Ed,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Don't forget that we (as members =
of the OMG)=20
ARE the OMG, in much the same way that we ARE Eclipse. The only =
chance we=20
have for improvement is for individuals to step up and make =
things=20
happen. I'm optimistic that the feedback we received at the first =
symposium=20
(and should receive at the second one in June) may be the catalyst =
needed for=20
change...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>In the meantime, I don't believe that =
this=20
particular issue needs to be raised to the OMG - see my =
response to=20
James, above.</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Kenn</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<BLOCKQUOTE=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
<DIV>"Ed Merks" <<A =
href=3D"mailto:merks@ca.ibm.com">merks@ca.ibm.com</A>>=20
wrote in message <A=20
=
href=3D"news:fsioea$3d6$1@build.eclipse.org">news:fsioea$3d6$1@build.ecli=
pse.org</A>...</DIV>Issue=20
tracking was one of the recurring themes at the EclipseCon OMG=20
symposium. <BR>
<BLOCKQUOTE><A=20
=
href=3D" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l">ht=
tp://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.html</A><BR></BLOC=
KQUOTE>If=20
issues aren't well tracked, they aren't likely to be well reported nor =
well=20
resolved, so it's in the OMG's best interest to=20
act...<BR><BR><BR>kaiserlautern wrote:=20
<BLOCKQUOTE cite=3Dmid:47ECB7B7.3000506@yahoo.fr type=3D"cite">Yes, =
you can. But=20
this is a typical chicken-and-egg problem, isn't it? ;-) =
<BR><BR>Dave=20
Carlson wrote: <BR>
<BLOCKQUOTE type=3D"cite">Last week at the Eclipse/OMG symposium, we =
had a=20
brief discussion that the OMG should move to a more modern issue =
tracking=20
system, like bugzilla. I find that the current OMG issue =
tracking is=20
so archaic that I just don't bother using it. I was =
considering this=20
same question earlier this morning as I uncovered several bugs in =
the OMG=20
cmof metamodel for SBVR.... <BR><BR>Can we submit an issue to OMG =
to=20
improve their issue tracking system :-) <BR><BR>Dave=20
<BR><BR>"kaiserlautern" <A class=3Dmoz-txt-link-rfc2396E=20
=
href=3D"mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></A>=
=20
wrote in message <A class=3Dmoz-txt-link-freetext=20
=
href=3D"news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</A=
>...=20
<BR>
<BLOCKQUOTE type=3D"cite">
<BLOCKQUOTE type=3D"cite">You may want to raise an RTF issue =
with the=20
OMG on this point. <BR></BLOCKQUOTE>Is there a reliable way of =
doing=20
this? Last time I sent a mail to <A =
class=3Dmoz-txt-link-abbreviated=20
href=3D"mailto:issues@omg.org">issues@omg.org</A>, they didn't =
even bother=20
sending me an automatic reply, let alone to file the issue.=20
=
<BR><BR></BLOCKQUOTE><BR><BR></BLOCKQUOTE></BLOCKQUOTE ><BR></BLOCKQUOTE><=
/BODY></HTML>
------=_NextPart_000_00C9_01C895AE.A72929C0--
|
|
|
| Re: OMG issue tracking [message #626349 is a reply to message #477188] |
Fri, 04 April 2008 11:19 |
Ed Merks
Messages: 33141
Registered: July 2009 |
Senior Member |
|
|
This is a multi-part message in MIME format.
--------------070002040706030708080501
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Kenn,
Really, I'm a member? And what exactly do you mean by implying that
whining isn't sufficient? It's so much easier than having to take
responsibility... :-P
Kenn Hussey wrote:
> Ed,
>
> Don't forget that we (as members of the OMG) ARE the OMG, in much the
> same way that we ARE Eclipse. The only chance we have for
> improvement is for individuals to step up and make things happen. I'm
> optimistic that the feedback we received at the first symposium (and
> should receive at the second one in June) may be the catalyst needed
> for change...
>
> In the meantime, I don't believe that this particular issue needs to
> be raised to the OMG - see my response to James, above.
>
> Kenn
>
>
> "Ed Merks" <merks@ca.ibm.com <mailto:merks@ca.ibm.com>> wrote in
> message news:fsioea$3d6$1@build.eclipse.org...
> Issue tracking was one of the recurring themes at the EclipseCon
> OMG symposium.
>
> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l
>
> If issues aren't well tracked, they aren't likely to be well
> reported nor well resolved, so it's in the OMG's best interest to
> act...
>
>
> kaiserlautern wrote:
>> Yes, you can. But this is a typical chicken-and-egg problem,
>> isn't it? ;-)
>>
>> Dave Carlson wrote:
>>> Last week at the Eclipse/OMG symposium, we had a brief
>>> discussion that the OMG should move to a more modern issue
>>> tracking system, like bugzilla. I find that the current OMG
>>> issue tracking is so archaic that I just don't bother using it.
>>> I was considering this same question earlier this morning as I
>>> uncovered several bugs in the OMG cmof metamodel for SBVR....
>>>
>>> Can we submit an issue to OMG to improve their issue tracking
>>> system :-)
>>>
>>> Dave
>>>
>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>> news:47EBC96A.1060008@yahoo.fr...
>>>>> You may want to raise an RTF issue with the OMG on this point.
>>>> Is there a reliable way of doing this? Last time I sent a mail
>>>> to issues@omg.org, they didn't even bother sending me an
>>>> automatic reply, let alone to file the issue.
>>>>
>>>
>>>
>
--------------070002040706030708080501
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body bgcolor="#ffffff" text="#000000">
Kenn,<br>
<br>
Really, I'm a member? And what exactly do you mean by implying that
whining isn't sufficient? It's so much easier than having to take
responsibility... :-P<br>
<br>
<br>
Kenn Hussey wrote:
<blockquote cite="mid:ft466a$bbn$3@build.eclipse.org" type="cite">
<meta http-equiv="Content-Type" content="text/html;charset=ISO-8859-1">
<meta content="MSHTML 6.00.6000.16608" name="GENERATOR">
<style></style>
<div><font face="Arial" size="2">Ed,</font></div>
<div> </div>
<div><font face="Arial" size="2">Don't forget that we (as members of
the OMG) ARE the OMG, in much the same way that we ARE Eclipse. The
only chance we have for improvement is for individuals to step up and
make things happen. I'm optimistic that the feedback we received at the
first symposium (and should receive at the second one in June) may be
the catalyst needed for change...</font></div>
<div> </div>
<div><font face="Arial" size="2">In the meantime, I don't believe
that this particular issue needs to be raised to the OMG - see my
response to James, above.</font></div>
<div> </div>
<div><font face="Arial" size="2">Kenn</font></div>
<div> </div>
<blockquote
style="border-left: 2px solid rgb(0, 0, 0); padding-right: 0px; padding-left: 5px; margin-left: 5px; margin-right: 0px;">
<div>"Ed Merks" <<a moz-do-not-send="true"
href="mailto:merks@ca.ibm.com">merks@ca.ibm.com</a>> wrote in
message <a moz-do-not-send="true"
href="news:fsioea$3d6$1@build.eclipse.org">news:fsioea$3d6$1@build.eclipse.org</a>...</div>
Issue tracking was one of the recurring themes at the EclipseCon OMG
symposium. <br>
<blockquote><a moz-do-not-send="true"
href=" http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l"> http://ed-merks.blogspot.com/2008/03/eclipsecon-thursday.htm l</a><br>
</blockquote>
If issues aren't well tracked, they aren't likely to be well reported
nor well resolved, so it's in the OMG's best interest to act...<br>
<br>
<br>
kaiserlautern wrote:
<blockquote cite="mid:47ECB7B7.3000506@yahoo.fr" type="cite">Yes,
you can. But this is a typical chicken-and-egg problem, isn't it? ;-) <br>
<br>
Dave Carlson wrote: <br>
<blockquote type="cite">Last week at the Eclipse/OMG symposium,
we had a brief discussion that the OMG should move to a more modern
issue tracking system, like bugzilla. I find that the current OMG
issue tracking is so archaic that I just don't bother using it. I was
considering this same question earlier this morning as I uncovered
several bugs in the OMG cmof metamodel for SBVR.... <br>
<br>
Can we submit an issue to OMG to improve their issue tracking system
:-) <br>
<br>
Dave <br>
<br>
"kaiserlautern" <a moz-do-not-send="true" class="moz-txt-link-rfc2396E"
href="mailto:comouraf-lixo@yahoo.fr"><comouraf-lixo@yahoo.fr></a>
wrote in message <a moz-do-not-send="true"
class="moz-txt-link-freetext" href="news:47EBC96A.1060008@yahoo.fr">news:47EBC96A.1060008@yahoo.fr</a>...
<br>
<blockquote type="cite">
<blockquote type="cite">You may want to raise an RTF issue
with the OMG on this point. <br>
</blockquote>
Is there a reliable way of doing this? Last time I sent a mail to <a
moz-do-not-send="true" class="moz-txt-link-abbreviated"
href="mailto:issues@omg.org">issues@omg.org</a>, they didn't even
bother sending me an automatic reply, let alone to file the issue. <br>
<br>
</blockquote>
<br>
<br>
</blockquote>
</blockquote>
<br>
</blockquote>
</blockquote>
<br>
</body>
</html>
--------------070002040706030708080501--
Ed Merks
Professional Support: https://www.macromodeling.com/
|
|
| | |
| Re: UML templates [message #626359 is a reply to message #477200] |
Fri, 04 April 2008 18:57 |
james bruck
Messages: 1724
Registered: July 2009 |
Senior Member |
|
|
Hi Kenn,
Say I had two Properties in a TemplateableElement, n and m. I want to say
that the type of n is "E" and the type of m is "E". So we have n:E and
m:E. E is a parameterable element of type Class owned by some template
parameter. When I then bind formal parameter "E" to some actual parameter
"Car" I would then have n:Car and m:Car (in the bound class).
In this example, we have one template parameter used in several places in
the template which is great, however, I wouldn't be able to do the same
thing with multiplicities since the ValueSpecification representing upper
and lower values would have to be owned by only one element. I think this
is a bit awkward.
... or am I missing something?
Cheers,
- James.
"Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
news:ft5guo$c0c$1@build.eclipse.org...
> James,
>
> You are correct, this it how it works - each entity (whether it be a type
> or a value specification) must be exposed using a separate template
> parameter. I don't personally see an issue with this...
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:ft5edu$n98$3@build.eclipse.org...
>> Hi Kenn,
>>
>> It is true that the TemplateParameter does not have to own its
>> ParameterableElements so the TemplateParameter could reference the
>> multiplicity value specification for a property, however, the problem
>> arises when we have two or more properties whose multiplicity should
>> change as a result of TemplateParameterSubstitution.
>> Since the ValueSpecification (ParameterableElement) representing the
>> multiplicity is directly owned by only one entity, we could not express
>> the fact that we want to substitute the multiplicities of many properties
>> with one template parameter.
>>
>> Contrast this with changing the Type of a property via
>> TemplateParameterSubstitution. A property references its Type (does not
>> directly own it) so it does not suffer from this issue. We can for
>> example have the TemplateParameter own the ParameterableElement
>> representing the type and then formulate multiple substitutions...no
>> problem.
>>
>>
>> ... Please correct me if I'm wrong.
>>
>> Cheers,
>>
>> - James.
>>
>>
>>
>>
>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>> news:ft4669$bbn$1@build.eclipse.org...
>>> Wait a minute. A template parameter need not own its parameterable
>>> elements... why don't you just reference the value specification via
>>> TemplateParameter::parameteredElement?
>>>
>>> Kenn
>>>
>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>> Hi C
|
|
|
| Re: UML templates [message #626360 is a reply to message #477202] |
Mon, 07 April 2008 13:38 |
Kenn Hussey
Messages: 1620
Registered: July 2009 |
Senior Member |
|
|
James,
Yes, it's a bit awkward, but that's the difference between copy by reference
and copy by value...
Kenn
"James Bruck" <jbruck@ca.ibm.com> wrote in message
news:ft5tmo$sgk$3@build.eclipse.org...
> Hi Kenn,
>
> Say I had two Properties in a TemplateableElement, n and m. I want to
> say that the type of n is "E" and the type of m is "E". So we have n:E
> and m:E. E is a parameterable element of type Class owned by some
> template parameter. When I then bind formal parameter "E" to some
> actual parameter "Car" I would then have n:Car and m:Car (in the bound
> class).
> In this example, we have one template parameter used in several places in
> the template which is great, however, I wouldn't be able to do the same
> thing with multiplicities since the ValueSpecification representing upper
> and lower values would have to be owned by only one element. I think
> this is a bit awkward.
>
> .. or am I missing something?
>
> Cheers,
> - James.
>
>
>
> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
> news:ft5guo$c0c$1@build.eclipse.org...
>> James,
>>
>> You are correct, this it how it works - each entity (whether it be a type
>> or a value specification) must be exposed using a separate template
>> parameter. I don't personally see an issue with this...
>>
>> Kenn
>>
>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>> news:ft5edu$n98$3@build.eclipse.org...
>>> Hi Kenn,
>>>
>>> It is true that the TemplateParameter does not have to own its
>>> ParameterableElements so the TemplateParameter could reference the
>>> multiplicity value specification for a property, however, the problem
>>> arises when we have two or more properties whose multiplicity should
>>> change as a result of TemplateParameterSubstitution.
>>> Since the ValueSpecification (ParameterableElement) representing the
>>> multiplicity is directly owned by only one entity, we could not express
>>> the fact that we want to substitute the multiplicities of many
>>> properties with one template parameter.
>>>
>>> Contrast this with changing the Type of a property via
>>> TemplateParameterSubstitution. A property references its Type (does not
>>> directly own it) so it does not suffer from this issue. We can for
>>> example have the TemplateParameter own the ParameterableElement
>>> representing the type and then formulate multiple substitutions...no
>>> problem.
>>>
>>>
>>> ... Please correct me if I'm wrong.
>>>
>>> Cheers,
>>>
>>> - James.
>>>
>>>
>>>
>>>
>>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>>> news:ft4669$bbn$1@build.eclipse.org...
>>>> Wait a minute. A template parameter need not own its parameterable
>>>> elements... why don't you just reference the value specification via
>>>> TemplateParameter::parameteredElement?
>>>>
>>>> Kenn
>>>>
>>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>>> Hi C
|
|
|
| Re: UML templates [message #626440 is a reply to message #477203] |
Thu, 24 April 2008 20:06 |
| Eclipse User |
|
|
|
Originally posted by: comouraf-lixo.yahoo.fr
Sorry for the delay. I've been busy for a while with the m2m (QVTO) forum
;-). In fact, our need in terms of parameterizations is very
straightforward: we usually have components with sometimes dozens of
multiplicityElements inside (ports, parts, connectorEnds) and we cannot
afford to hang dozens of parameters to the same signature when - usually -
all multiplicities should change synchronously. In this case, a single
parameter is, most of the time, enough. As a workaround, I created a
NamedParameter stereotype so that internal elements can refer to parameters
by their names, but ideally, it would be better if MultiplicityElement had
also simple references to ValueSpecification (in addition to the
aggregation-type lowerValue and upperValue). By the way, did anybody find a
good reason for a TemplateParameter not to extend NamedElement? ;-)
César
"Kenn Hussey" <Kenn.Hussey@embarcadero.com> a écrit dans le message de
news:ftd84j$35a$1@build.eclipse.org...
> James,
>
> Yes, it's a bit awkward, but that's the difference between copy by
> reference and copy by value...
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:ft5tmo$sgk$3@build.eclipse.org...
>> Hi Kenn,
>>
>> Say I had two Properties in a TemplateableElement, n and m. I want to
>> say that the type of n is "E" and the type of m is "E". So we have n:E
>> and m:E. E is a parameterable element of type Class owned by some
>> template parameter. When I then bind formal parameter "E" to some
>> actual parameter "Car" I would then have n:Car and m:Car (in the bound
>> class).
>> In this example, we have one template parameter used in several places in
>> the template which is great, however, I wouldn't be able to do the same
>> thing with multiplicities since the ValueSpecification representing upper
>> and lower values would have to be owned by only one element. I think
>> this is a bit awkward.
>>
>> .. or am I missing something?
>>
>> Cheers,
>> - James.
>>
>>
>>
>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>> news:ft5guo$c0c$1@build.eclipse.org...
>>> James,
>>>
>>> You are correct, this it how it works - each entity (whether it be a
>>> type or a value specification) must be exposed using a separate template
>>> parameter. I don't personally see an issue with this...
>>>
>>> Kenn
>>>
>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>> news:ft5edu$n98$3@build.eclipse.org...
>>>> Hi Kenn,
>>>>
>>>> It is true that the TemplateParameter does not have to own its
>>>> ParameterableElements so the TemplateParameter could reference the
>>>> multiplicity value specification for a property, however, the problem
>>>> arises when we have two or more properties whose multiplicity should
>>>> change as a result of TemplateParameterSubstitution.
>>>> Since the ValueSpecification (ParameterableElement) representing the
>>>> multiplicity is directly owned by only one entity, we could not express
>>>> the fact that we want to substitute the multiplicities of many
>>>> properties with one template parameter.
>>>>
>>>> Contrast this with changing the Type of a property via
>>>> TemplateParameterSubstitution. A property references its Type (does
>>>> not directly own it) so it does not suffer from this issue. We can for
>>>> example have the TemplateParameter own the ParameterableElement
>>>> representing the type and then formulate multiple substitutions...no
>>>> problem.
>>>>
>>>>
>>>> ... Please correct me if I'm wrong.
>>>>
>>>> Cheers,
>>>>
>>>> - James.
>>>>
>>>>
>>>>
>>>>
>>>> "Kenn Hussey" <Kenn.Hussey@embarcadero.com> wrote in message
>>>> news:ft4669$bbn$1@build.eclipse.org...
>>>>> Wait a minute. A template parameter need not own its parameterable
>>>>> elements... why don't you just reference the value specification via
>>>>> TemplateParameter::parameteredElement?
>>>>>
>>>>> Kenn
>>>>>
>>>>> "James Bruck" <jbruck@ca.ibm.com> wrote in message
>>>>> news:fsg7ki$coc$2@build.eclipse.org...
>>>>>> Hi César,
>>>>>>
>>>>>> The second posting clarified it for me. I see the issue.
>>>>>> Unfortunately it appears that you will not be able to reuse your
>>>>>> value specification.
>>>>>> As you mentioned, an element cannot be owned by more than one
>>>>>> container. This seems to be a basic problem that UML should address
>>>>>> or at the very least it should be brought to the attention of those
>>>>>> defining the spec.
>>>>>>
>>>>>> You may want to raise an RTF issue with the OMG on this point.
>>>>>>
>>>>>> Cheers,
>>>>>> - James.
>>>>>>
>>>>>>
>>>>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>>>>> news:47EB85BA.4050409@yahoo.fr...
>>>>>>> Hi, James,
>>>>>>>
>>>>>>> Thanks for your answer. By the way, I had already read your good
>>>>>>> article on generics.
>>>>>>>
>>>>>>> I had also looked for related questions in the UML issues, but to no
>>>>>>> avail. In fact what I need to "expose" is not the property
>>>>>>> (MultiplicityElement) but its multiplicity (i.e. a
>>>>>>> ValueSpecification). And according to the UML spec, value
>>>>>>> specifications can be exposed. As I said before, a parameter can
>>>>>>> expose only one multiplicity value while I needed to have the same
>>>>>>> parameter exposing more than one multiplicity value.
>>>>>>>
>>>>>>> To solve this problem, I use the aggregation type
>>>>>>> ownedParametedElement (i.e. the parameterable element belongs to the
>>>>>>> template parameter). And this parameterable element, in this case,
>>>>>>> is my value specification (say, vs1) that will be used as a value
>>>>>>> for my exposed multiplicities. However, when I specify that my
>>>>>>> property1 has a multiplicity value (upperValue) equals to vs1, it
>>>>>>> takes possession of it and I cannot reuse vs1 as an upperValue for
>>>>>>> another property (because an element cannot be owned by more than
>>>>>>> one "container"). And this contrasts with the relationship between a
>>>>>>> property and its type, which is just a simple reference (that's why
>>>>>>> in my simple example we can have more than one property with the
>>>>>>> same type T)
>>>>>>>
>>>>>>> Schematically, we would have (please make an extra effort to
>>>>>>> understand these diagrams ;-)
>>>>>>>
>>>>>>> ------------------------
>>>>>>> ParameterElement
>>>>>>> ------------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ParameterableElement
>>>>>>> ----------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ValueSpecification
>>>>>>> ----------------------
>>>>>>> ^
>>>>>>> | (aggregation)
>>>>>>> <>
>>>>>>> ------------------------
>>>>>>> MultiplicityElement
>>>>>>> -----------------------
>>>>>>>
>>>>>>> ---------------------------------------------------------
>>>>>>>
>>>>>>> ------------------------
>>>>>>> ParameterElement
>>>>>>> ------------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> ParameterableElement
>>>>>>> ----------------------
>>>>>>> <>
>>>>>>> |
>>>>>>> v
>>>>>>> -----------------------
>>>>>>> Classifier
>>>>>>> ----------------------
>>>>>>> ^ ^
>>>>>>> | | (simple references)
>>>>>>> | |
>>>>>>> --------------- --------------
>>>>>>> Property1 Property2
>>>>>>> --------------- --------------
>>>>>>>
>>>>>>>
>>>>>>> Hope my problem is better explained now.
>>>>>>>
>>>>>>> César
>>>>>>>
>>>>>>> James Bruck wrote:
>>>>>>>> Hi César,
>>>>>>>>
>>>>>>>> The "exposing" of elements is handled by the TemplateParameter
>>>>>>>> type. From the spec "...A TemplateParameter exposes
>>>>>>>> ParameterableElements as formal parameters...".
>>>>>>>> If I understand your question correctly, you want to specify the
>>>>>>>> multiplicity of the property as a parameter. Unfortunately it
>>>>>>>> appears that MultiplicityElement is not a ParameterableElement.
>>>>>>>> That would seem to imply that one would not be able to specify
>>>>>>>> TemplateParameterSubstitutions for some TemplateBinding to modify
>>>>>>>> the bounds of the property.
>>>>>>>> I took a quick look at all the RTF issues currently logged and this
>>>>>>>> did not appear to be an issue that has been raised.
>>>>>>>>
>>>>>>>> The following might be of some use...
>>>>>>>> http://www.eclipse.org/modeling/mdt/uml2/docs/articles/Defin ing_Generics_with_UML_Templates/article.html
>>>>>>>>
>>>>>>>> I will dig more deeply into this tomorrow.
>>>>>>>>
>>>>>>>> Cheers,
>>>>>>>> - James.
>>>>>>>>
>>>>>>>> "kaiserlautern" <comouraf-lixo@yahoo.fr> wrote in message
>>>>>>>> news:fsdbha$b4i$1@build.eclipse.org...
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> I have a problem with using templates and don't know if it is
>>>>>>>>> caused by a real issue from the UML spec or if it is something I
>>>>>>>>> missed.
>>>>>>>>>
>>>>>>>>> A very frequent situation in our component-based modeling process
>>>>>>>>> cannot be properly addressed by the UML templating mechanism as it
>>>>>>>>> is now. The problem we face is similar to the one illustrated in
>>>>>>>>> the figure below and concerns the fact that not always can the
>>>>>>>>> same parameter "expose" (to use the UML lingo) two or more
>>>>>>>>> elements.
>>>>>>>>>
>>>>>>>>> ClassA<T, n>
>>>>>>>>> prop1: T[n]
>>>>>>>>> prop2: T[n]
>>>>>>>>>
>>>>>>>>> The example above works for parameter T, while it fails with n.
>>>>>>>>> Yet, the need of having multiple elements whose multiplicity is
>>>>>>>>> parameterized by a same parameter seems quite plausible. Taking a
>>>>>>>>> look at the UML metamodel we can see that the references between a
>>>>>>>>> property and its type, on one side, and between a property and its
>>>>>>>>> upper bound multiplicity (i.e. upperValue), on the other side,
>>>>>>>>> present different natures --the first case being a simple
>>>>>>>>> reference and the second, an aggregation--, what makes all the
>>>>>>>>> difference when you want to expose the same element multiple
>>>>>>>>> times. In our modeling process, most often than not, a given
>>>>>>>>> parameter needs to be used by more than one element inside a
>>>>>>>>> template.
>>>>>>>>>
>>>>>>>>> Can anyone comment on this?
>>>>>>>>>
>>>>>>>>> Thanks in advance
>>>>>>>>>
>>>>>>>>> César
>>>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>
|
|
| |
Goto Forum:
Current Time: Thu Apr 25 19:44:04 GMT 2024
Powered by FUDForum. Page generated in 0.07576 seconds |
] 
Search
Help
Register
Login
Home
