Home » Modeling » UML2 » heavyweight extension in OMG specifications
| |
Re: heavyweight extension in OMG specifications [message #473094 is a reply to message #473092] |
Tue, 05 June 2007 03:34 |
Kenn Hussey Messages: 1620 Registered: July 2009 |
Senior Member |
|
|
Actually, I believe the UML and/or MOF specification does use the term
"heavyweight" when referring to extensions built by merging one or more
packages from UML (InfrastructureLibrary) - I certainly didn't invent this
term.
Kenn
"James Bruck" <jbruck@ca.ibm.com> wrote in message
news:f41r75$780$2@build.eclipse.org...
> Hi Manar,
>
> The heavyweight and middleweight extension mechanisms are not explicitly
> mentioned in the MOF specification however, the notion of using package
> merge to define languages is mentioned and forms the bases of heavyweight
> and middleweight extensions.
>
> UML itself is built using the package merge concept and extensions can
> also
> be achieved with the same technique.
>
> By the way, the terminology "Middleweight" and "Heavyweight" are terms
> coined by Kenn Hussey and are not really official OMG terminology. We try
> to be consistent when using them in documentation etc. The document in
> the
> "Improved Documentation" bugzilla uses a slightly different term but will
> be
> updated.
>
> Package import and merge are somewhat key to these kinds of extension
> mechanisms.
>
> Regards,
>
> - James.
>
>
>
>
> "Manar El-Kady" <manar.elkady@gmail.com> wrote in message
> news:f3u0mr$q90$1@build.eclipse.org...
>> Hi,
>>
>> Are both of the heavyweight and the middleweight extensions that you
>> illustrated are considered the (first-class extensibility)heavyweight
>> extension mechanism that specified by the UML specification?
>>
>> In many articles and papers, it refers to the heavyweight extension that
>> was defined by the MOF. I opened the MOF specification and I didn't
>> any mention to heavyweight or first-class extensibility. The only thing
>> that I found is Extension package, which I didn't understand its usage.
>> Do you clear this to me?
>>
>> Are the package import and emerge considered essential steps while
>> applying the heavyweight extension?
>>
>>
>> Thanks in advance,
>>
>> Manar
>
>
|
|
|
Re: heavyweight extension in OMG specifications [message #473188 is a reply to message #473094] |
Tue, 12 June 2007 15:42 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
I think the terms "Middleweight" and "Flyweight" are not mentioned in the
spec. Although the use of keywords and package merge are mentioned.
You gotta take credit for something Kenn :)
- James.
"Kenn Hussey" <khussey@ca.ibm.com> wrote in message
news:f42lk8$6at$1@build.eclipse.org...
> Actually, I believe the UML and/or MOF specification does use the term
> "heavyweight" when referring to extensions built by merging one or more
> packages from UML (InfrastructureLibrary) - I certainly didn't invent this
> term.
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:f41r75$780$2@build.eclipse.org...
> > Hi Manar,
> >
> > The heavyweight and middleweight extension mechanisms are not explicitly
> > mentioned in the MOF specification however, the notion of using package
> > merge to define languages is mentioned and forms the bases of
heavyweight
> > and middleweight extensions.
> >
> > UML itself is built using the package merge concept and extensions can
> > also
> > be achieved with the same technique.
> >
> > By the way, the terminology "Middleweight" and "Heavyweight" are terms
> > coined by Kenn Hussey and are not really official OMG terminology. We
try
> > to be consistent when using them in documentation etc. The document in
> > the
> > "Improved Documentation" bugzilla uses a slightly different term but
will
> > be
> > updated.
> >
> > Package import and merge are somewhat key to these kinds of extension
> > mechanisms.
> >
> > Regards,
> >
> > - James.
> >
> >
> >
> >
> > "Manar El-Kady" <manar.elkady@gmail.com> wrote in message
> > news:f3u0mr$q90$1@build.eclipse.org...
> >> Hi,
> >>
> >> Are both of the heavyweight and the middleweight extensions that you
> >> illustrated are considered the (first-class extensibility)heavyweight
> >> extension mechanism that specified by the UML specification?
> >>
> >> In many articles and papers, it refers to the heavyweight extension
that
> >> was defined by the MOF. I opened the MOF specification and I didn't
> >> any mention to heavyweight or first-class extensibility. The only thing
> >> that I found is Extension package, which I didn't understand its usage.
> >> Do you clear this to me?
> >>
> >> Are the package import and emerge considered essential steps while
> >> applying the heavyweight extension?
> >>
> >>
> >> Thanks in advance,
> >>
> >> Manar
> >
> >
>
>
|
|
| |
Re: heavyweight extension in OMG specifications [message #473190 is a reply to message #473189] |
Tue, 12 June 2007 17:00 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
Hi Dave,
Keywords are mentioned in Annex B of the spec.
They do have a non-standard representation and are saved as annotations as
you mentioned. Therefore, if you decide to export your model to XMI, the
annotations will be moved into XMI extensions.
Keywords have limited functionality, they are just a way to visually
distinguish between similar looking items. There is no concept of a
dictionary to share with others, no way to validate the appplication of
keywords etc.
There is support in the API with Element#addKeyword, Element#getKeyword()
Element#removeKeyword().
Cheers,
- James.
"Dave Carlson" <dcarlson@xmlmodeling.com> wrote in message
news:f4mgul$9ck$1@build.eclipse.org...
> Hi James,
> Where and how are use of keywords mentioned in the spec? I've had the
> concern that keywords are not part of the standard metamodel, and
> therefore not portable across UML tools. The UML2 model implements
> these as EMF annotations, or XMI extensions.
>
> Although keywords appear to be quite useful, I try to avoid them due to
> non-standard representation.
>
> Dave
>
> James Bruck wrote:
> > I think the terms "Middleweight" and "Flyweight" are not mentioned in
the
> > spec. Although the use of keywords and package merge are mentioned.
> > You gotta take credit for something Kenn :)
> >
|
|
|
Re: heavyweight extension in OMG specifications [message #473191 is a reply to message #473190] |
Tue, 12 June 2007 17:42 |
Dave Carlson Messages: 402 Registered: July 2009 |
Senior Member |
|
|
Thanks, James. Yes, I've used the API support for both setting and
reading keywords. The greatest confusion among modeling users is when
working with RSM/A, where there is a nice Keywords field right on the
Stereotypes tab... So, especially when users have worked with Rose,
they see the Keyword as synonymous with a stereotype (Rose had a very
relaxed idea of stereotypes, where you could just type anything into the
Stereotype field).
So, anyway, my question relates more to to what kind of advice I should
give these users. I generally tell them to avoid using keywords unless,
as you say, they only want to visually distinguish model elements.
I'll check Annex B as a reference that I can pass on to others.
Thanks,
Dave
James Bruck wrote:
> Hi Dave,
>
> Keywords are mentioned in Annex B of the spec.
>
> They do have a non-standard representation and are saved as annotations as
> you mentioned. Therefore, if you decide to export your model to XMI, the
> annotations will be moved into XMI extensions.
> Keywords have limited functionality, they are just a way to visually
> distinguish between similar looking items. There is no concept of a
> dictionary to share with others, no way to validate the appplication of
> keywords etc.
>
> There is support in the API with Element#addKeyword, Element#getKeyword()
> Element#removeKeyword().
>
|
|
|
Re: heavyweight extension in OMG specifications [message #473197 is a reply to message #473191] |
Wed, 13 June 2007 13:47 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
Hi Dave,
Apart from just explaining the differences between keywords and stereotypes
and pointing them to Annex B of the UML superstructure spec.
http://www.omg.org/docs/ptc/06-04-02.pdf or later ( there is a 07-02-05
version), I don't think I would have other advice.
The use of Keywords and Stereotypes in UML 1.4 ( as Rose was using ) and
2.1.1 is significantly different and users have to understand that.
One other resource would be an article that will be published soon called
"Customizing UML: Which technique is right for you?" It will explain the
use of keywords/profiles to customize UML. It is currently in draft form
attached to a bugzilla called "Improved Documentation".
I'm not sure if RSM/A has a help topic on this but perhaps you could suggest
it to them if this is something you would really like to see.
Cheers,
- James.
"Dave Carlson" <dcarlson@xmlmodeling.com> wrote in message
news:f4mlte$ojq$1@build.eclipse.org...
> Thanks, James. Yes, I've used the API support for both setting and
> reading keywords. The greatest confusion among modeling users is when
> working with RSM/A, where there is a nice Keywords field right on the
> Stereotypes tab... So, especially when users have worked with Rose,
> they see the Keyword as synonymous with a stereotype (Rose had a very
> relaxed idea of stereotypes, where you could just type anything into the
> Stereotype field).
>
> So, anyway, my question relates more to to what kind of advice I should
> give these users. I generally tell them to avoid using keywords unless,
> as you say, they only want to visually distinguish model elements.
>
> I'll check Annex B as a reference that I can pass on to others.
>
> Thanks,
> Dave
>
> James Bruck wrote:
> > Hi Dave,
> >
> > Keywords are mentioned in Annex B of the spec.
> >
> > They do have a non-standard representation and are saved as annotations
as
> > you mentioned. Therefore, if you decide to export your model to XMI,
the
> > annotations will be moved into XMI extensions.
> > Keywords have limited functionality, they are just a way to visually
> > distinguish between similar looking items. There is no concept of a
> > dictionary to share with others, no way to validate the appplication of
> > keywords etc.
> >
> > There is support in the API with Element#addKeyword,
Element#getKeyword()
> > Element#removeKeyword().
> >
|
|
| |
Re: heavyweight extension in OMG specifications [message #619682 is a reply to message #473092] |
Tue, 05 June 2007 03:34 |
Kenn Hussey Messages: 1620 Registered: July 2009 |
Senior Member |
|
|
Actually, I believe the UML and/or MOF specification does use the term
"heavyweight" when referring to extensions built by merging one or more
packages from UML (InfrastructureLibrary) - I certainly didn't invent this
term.
Kenn
"James Bruck" <jbruck@ca.ibm.com> wrote in message
news:f41r75$780$2@build.eclipse.org...
> Hi Manar,
>
> The heavyweight and middleweight extension mechanisms are not explicitly
> mentioned in the MOF specification however, the notion of using package
> merge to define languages is mentioned and forms the bases of heavyweight
> and middleweight extensions.
>
> UML itself is built using the package merge concept and extensions can
> also
> be achieved with the same technique.
>
> By the way, the terminology "Middleweight" and "Heavyweight" are terms
> coined by Kenn Hussey and are not really official OMG terminology. We try
> to be consistent when using them in documentation etc. The document in
> the
> "Improved Documentation" bugzilla uses a slightly different term but will
> be
> updated.
>
> Package import and merge are somewhat key to these kinds of extension
> mechanisms.
>
> Regards,
>
> - James.
>
>
>
>
> "Manar El-Kady" <manar.elkady@gmail.com> wrote in message
> news:f3u0mr$q90$1@build.eclipse.org...
>> Hi,
>>
>> Are both of the heavyweight and the middleweight extensions that you
>> illustrated are considered the (first-class extensibility)heavyweight
>> extension mechanism that specified by the UML specification?
>>
>> In many articles and papers, it refers to the heavyweight extension that
>> was defined by the MOF. I opened the MOF specification and I didn't
>> any mention to heavyweight or first-class extensibility. The only thing
>> that I found is Extension package, which I didn't understand its usage.
>> Do you clear this to me?
>>
>> Are the package import and emerge considered essential steps while
>> applying the heavyweight extension?
>>
>>
>> Thanks in advance,
>>
>> Manar
>
>
|
|
|
Re: heavyweight extension in OMG specifications [message #621556 is a reply to message #473094] |
Tue, 12 June 2007 15:42 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
I think the terms "Middleweight" and "Flyweight" are not mentioned in the
spec. Although the use of keywords and package merge are mentioned.
You gotta take credit for something Kenn :)
- James.
"Kenn Hussey" <khussey@ca.ibm.com> wrote in message
news:f42lk8$6at$1@build.eclipse.org...
> Actually, I believe the UML and/or MOF specification does use the term
> "heavyweight" when referring to extensions built by merging one or more
> packages from UML (InfrastructureLibrary) - I certainly didn't invent this
> term.
>
> Kenn
>
> "James Bruck" <jbruck@ca.ibm.com> wrote in message
> news:f41r75$780$2@build.eclipse.org...
> > Hi Manar,
> >
> > The heavyweight and middleweight extension mechanisms are not explicitly
> > mentioned in the MOF specification however, the notion of using package
> > merge to define languages is mentioned and forms the bases of
heavyweight
> > and middleweight extensions.
> >
> > UML itself is built using the package merge concept and extensions can
> > also
> > be achieved with the same technique.
> >
> > By the way, the terminology "Middleweight" and "Heavyweight" are terms
> > coined by Kenn Hussey and are not really official OMG terminology. We
try
> > to be consistent when using them in documentation etc. The document in
> > the
> > "Improved Documentation" bugzilla uses a slightly different term but
will
> > be
> > updated.
> >
> > Package import and merge are somewhat key to these kinds of extension
> > mechanisms.
> >
> > Regards,
> >
> > - James.
> >
> >
> >
> >
> > "Manar El-Kady" <manar.elkady@gmail.com> wrote in message
> > news:f3u0mr$q90$1@build.eclipse.org...
> >> Hi,
> >>
> >> Are both of the heavyweight and the middleweight extensions that you
> >> illustrated are considered the (first-class extensibility)heavyweight
> >> extension mechanism that specified by the UML specification?
> >>
> >> In many articles and papers, it refers to the heavyweight extension
that
> >> was defined by the MOF. I opened the MOF specification and I didn't
> >> any mention to heavyweight or first-class extensibility. The only thing
> >> that I found is Extension package, which I didn't understand its usage.
> >> Do you clear this to me?
> >>
> >> Are the package import and emerge considered essential steps while
> >> applying the heavyweight extension?
> >>
> >>
> >> Thanks in advance,
> >>
> >> Manar
> >
> >
>
>
|
|
| |
Re: heavyweight extension in OMG specifications [message #621561 is a reply to message #473189] |
Tue, 12 June 2007 17:00 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
Hi Dave,
Keywords are mentioned in Annex B of the spec.
They do have a non-standard representation and are saved as annotations as
you mentioned. Therefore, if you decide to export your model to XMI, the
annotations will be moved into XMI extensions.
Keywords have limited functionality, they are just a way to visually
distinguish between similar looking items. There is no concept of a
dictionary to share with others, no way to validate the appplication of
keywords etc.
There is support in the API with Element#addKeyword, Element#getKeyword()
Element#removeKeyword().
Cheers,
- James.
"Dave Carlson" <dcarlson@xmlmodeling.com> wrote in message
news:f4mgul$9ck$1@build.eclipse.org...
> Hi James,
> Where and how are use of keywords mentioned in the spec? I've had the
> concern that keywords are not part of the standard metamodel, and
> therefore not portable across UML tools. The UML2 model implements
> these as EMF annotations, or XMI extensions.
>
> Although keywords appear to be quite useful, I try to avoid them due to
> non-standard representation.
>
> Dave
>
> James Bruck wrote:
> > I think the terms "Middleweight" and "Flyweight" are not mentioned in
the
> > spec. Although the use of keywords and package merge are mentioned.
> > You gotta take credit for something Kenn :)
> >
|
|
|
Re: heavyweight extension in OMG specifications [message #621564 is a reply to message #473190] |
Tue, 12 June 2007 17:42 |
Dave Carlson Messages: 402 Registered: July 2009 |
Senior Member |
|
|
Thanks, James. Yes, I've used the API support for both setting and
reading keywords. The greatest confusion among modeling users is when
working with RSM/A, where there is a nice Keywords field right on the
Stereotypes tab... So, especially when users have worked with Rose,
they see the Keyword as synonymous with a stereotype (Rose had a very
relaxed idea of stereotypes, where you could just type anything into the
Stereotype field).
So, anyway, my question relates more to to what kind of advice I should
give these users. I generally tell them to avoid using keywords unless,
as you say, they only want to visually distinguish model elements.
I'll check Annex B as a reference that I can pass on to others.
Thanks,
Dave
James Bruck wrote:
> Hi Dave,
>
> Keywords are mentioned in Annex B of the spec.
>
> They do have a non-standard representation and are saved as annotations as
> you mentioned. Therefore, if you decide to export your model to XMI, the
> annotations will be moved into XMI extensions.
> Keywords have limited functionality, they are just a way to visually
> distinguish between similar looking items. There is no concept of a
> dictionary to share with others, no way to validate the appplication of
> keywords etc.
>
> There is support in the API with Element#addKeyword, Element#getKeyword()
> Element#removeKeyword().
>
|
|
|
Re: heavyweight extension in OMG specifications [message #621578 is a reply to message #473191] |
Wed, 13 June 2007 13:47 |
james bruck Messages: 1724 Registered: July 2009 |
Senior Member |
|
|
Hi Dave,
Apart from just explaining the differences between keywords and stereotypes
and pointing them to Annex B of the UML superstructure spec.
http://www.omg.org/docs/ptc/06-04-02.pdf or later ( there is a 07-02-05
version), I don't think I would have other advice.
The use of Keywords and Stereotypes in UML 1.4 ( as Rose was using ) and
2.1.1 is significantly different and users have to understand that.
One other resource would be an article that will be published soon called
"Customizing UML: Which technique is right for you?" It will explain the
use of keywords/profiles to customize UML. It is currently in draft form
attached to a bugzilla called "Improved Documentation".
I'm not sure if RSM/A has a help topic on this but perhaps you could suggest
it to them if this is something you would really like to see.
Cheers,
- James.
"Dave Carlson" <dcarlson@xmlmodeling.com> wrote in message
news:f4mlte$ojq$1@build.eclipse.org...
> Thanks, James. Yes, I've used the API support for both setting and
> reading keywords. The greatest confusion among modeling users is when
> working with RSM/A, where there is a nice Keywords field right on the
> Stereotypes tab... So, especially when users have worked with Rose,
> they see the Keyword as synonymous with a stereotype (Rose had a very
> relaxed idea of stereotypes, where you could just type anything into the
> Stereotype field).
>
> So, anyway, my question relates more to to what kind of advice I should
> give these users. I generally tell them to avoid using keywords unless,
> as you say, they only want to visually distinguish model elements.
>
> I'll check Annex B as a reference that I can pass on to others.
>
> Thanks,
> Dave
>
> James Bruck wrote:
> > Hi Dave,
> >
> > Keywords are mentioned in Annex B of the spec.
> >
> > They do have a non-standard representation and are saved as annotations
as
> > you mentioned. Therefore, if you decide to export your model to XMI,
the
> > annotations will be moved into XMI extensions.
> > Keywords have limited functionality, they are just a way to visually
> > distinguish between similar looking items. There is no concept of a
> > dictionary to share with others, no way to validate the appplication of
> > keywords etc.
> >
> > There is support in the API with Element#addKeyword,
Element#getKeyword()
> > Element#removeKeyword().
> >
|
|
|
Goto Forum:
Current Time: Fri Sep 20 17:20:16 GMT 2024
Powered by FUDForum. Page generated in 0.04951 seconds
|