Home » Modeling » UML2 » Interaction Messages
|
| Re: Interaction Messages [message #474395 is a reply to message #474393] |
Fri, 20 July 2007 15:57   |
 james bruck
Messages: 1724
Registered: July 2009 |
Senior Member |
|
|
Hi Andrew,
I'll work on a more detailed answer to this question but as a quick note I
wanted to mention that if you are hoping to do code generation from sequence
diagrams solely, I don't think that you will be successfull because of
issues with sequence diagrams ( I would encourage others to aggree/disagree
with me in this newsgroup ).
I think that ultimately you will be forced to use Activity diagrams to fully
express the details required to generate code. You can of course use a
combination of Sequence/Activity to express details for code generation (in
fact, I would think that Class/Structure/Activity/Statechart/Sequence would
all be required since each describe different aspects of code
Structure/Behavior).
It's been my experience that sequence diagrams are great for sketching but
sometimes lack ability to express minute details.
In this case, I don't think it's the UML API that is holding us back but
rather an interpretation of the spec. If we can figure this out from the
UML spec, implementing this with the UML API should be straight forward.
As I mentioned above, I will provide a more detailed answer as soon as I
work out all the details. In the meantime, I would suggest considering
Activity diagrams for code generation.
- James.
"Andrew Carton" <cartona@cs.tcd.ie> wrote in message
news:f7qg9i$qve$1@build.eclipse.org...
> I'm trying to understand how to form valid sequence diagrams for code
> generation.
>
> I read the previous message on messages with return value and local
> variables but I still don't quite understand it.
>
> My first question is, how to pass the below as arguments to a message.
> As I understand it you can only create a ValueSpecification (and its
> subclasses) as an argument to a message (with its signature as an
> operation).
>
> The specification says you can pass these as values to the message.
>
> i) attributes of the sending lifeline.
> ii) constants.
> iii) symbolic values (which are wildcard values representing any legal
> value).
> iv) explicit parameters of the enclosing Interaction.
> v) attributes of the class owning the Interaction.
>
> For the first case, what exactly is an attribute of a sending lifeline
> and how do you create and reference it as a ValueSpecification? How do I
> reference as an argument an explicit parameter or an attribute of a
> class owning the Interaction.
>
> For return values, a previous post said to create an
> InstanceSpecification that refers to a ValueSpecification (I am assuming
> an InstanceValue here). The InstanceValue can then reference the
> InstanceSpecification and you can set the name as the variablename and
> the type as the type of the local variable. However, an
> InstanceSpecification is not really local as it belongs to a package and
> not an enclosing interaction or even the behaviourClassifier that the
> interaction belongs to. So this looks like it is a global instance?
> Also, how do you assign the return value to a class attribute that the
> lifeline represents?
>
> If someone has figured this out and can provide code I would really
> appreciate it as I am finding it quite difficult to understand how these
> are accomplished from both the specification and the uml2 api.
>
> Thanks,
> Andrew.
>
>
>
>
|
|
|
| Re: Interaction Messages [message #474507 is a reply to message #474395] |
Mon, 23 July 2007 15:40 |
Andrew Carton
Messages: 104
Registered: July 2009 |
Senior Member |
|
|
Hi James,
Thanks for answering. I understand that sequence diagrams are not
suitable to model the entire behaviour of a system, but I only hope to
generate code with the capabilities that they do offer. Any assistance
you could give on this I would greatly appreciate. I think you mentioned
in a previous post that there is a workaround by attaching expressions
to messages - and I am thinking of doing this - if there is no other way.
Thanks,
Andrew.
James Bruck wrote:
> Hi Andrew,
>
> I'll work on a more detailed answer to this question but as a quick note I
> wanted to mention that if you are hoping to do code generation from sequence
> diagrams solely, I don't think that you will be successfull because of
> issues with sequence diagrams ( I would encourage others to aggree/disagree
> with me in this newsgroup ).
>
> I think that ultimately you will be forced to use Activity diagrams to fully
> express the details required to generate code. You can of course use a
> combination of Sequence/Activity to express details for code generation (in
> fact, I would think that Class/Structure/Activity/Statechart/Sequence would
> all be required since each describe different aspects of code
> Structure/Behavior).
>
> It's been my experience that sequence diagrams are great for sketching but
> sometimes lack ability to express minute details.
>
> In this case, I don't think it's the UML API that is holding us back but
> rather an interpretation of the spec. If we can figure this out from the
> UML spec, implementing this with the UML API should be straight forward.
>
> As I mentioned above, I will provide a more detailed answer as soon as I
> work out all the details. In the meantime, I would suggest considering
> Activity diagrams for code generation.
>
> - James.
>
>
> "Andrew Carton" <cartona@cs.tcd.ie> wrote in message
> news:f7qg9i$qve$1@build.eclipse.org...
>> I'm trying to understand how to form valid sequence diagrams for code
>> generation.
>>
>> I read the previous message on messages with return value and local
>> variables but I still don't quite understand it.
>>
>> My first question is, how to pass the below as arguments to a message.
>> As I understand it you can only create a ValueSpecification (and its
>> subclasses) as an argument to a message (with its signature as an
>> operation).
>>
>> The specification says you can pass these as values to the message.
>>
>> i) attributes of the sending lifeline.
>> ii) constants.
>> iii) symbolic values (which are wildcard values representing any legal
>> value).
>> iv) explicit parameters of the enclosing Interaction.
>> v) attributes of the class owning the Interaction.
>>
>> For the first case, what exactly is an attribute of a sending lifeline
>> and how do you create and reference it as a ValueSpecification? How do I
>> reference as an argument an explicit parameter or an attribute of a
>> class owning the Interaction.
>>
>> For return values, a previous post said to create an
>> InstanceSpecification that refers to a ValueSpecification (I am assuming
>> an InstanceValue here). The InstanceValue can then reference the
>> InstanceSpecification and you can set the name as the variablename and
>> the type as the type of the local variable. However, an
>> InstanceSpecification is not really local as it belongs to a package and
>> not an enclosing interaction or even the behaviourClassifier that the
>> interaction belongs to. So this looks like it is a global instance?
>> Also, how do you assign the return value to a class attribute that the
>> lifeline represents?
>>
>> If someone has figured this out and can provide code I would really
>> appreciate it as I am finding it quite difficult to understand how these
>> are accomplished from both the specification and the uml2 api.
>>
>> Thanks,
>> Andrew.
>>
>>
>>
>>
>
>
|
|
|
| Re: Interaction Messages [message #474509 is a reply to message #474507] |
Tue, 24 July 2007 11:00 |
Andrew Carton
Messages: 104
Registered: July 2009 |
Senior Member |
|
|
Hey,
If anyone else is looking at sequence diagrams, I have found some
unresolved issues at the OMG's UML2 RTF site relating to storing and
sending arguments with messages. The ones of particular interest:
http://www.omg.org/issues/uml2-rtf.open.html#Issue8785
http://www.omg.org/issues/uml2-rtf.open.html#Issue8786
http://www.omg.org/issues/uml2-rtf.open.html#Issue8787
http://www.omg.org/issues/uml2-rtf.open.html#Issue8788
http://www.omg.org/issues/uml2-rtf.open.html#Issue8899
http://www.omg.org/issues/uml2-rtf.open.html#Issue10591
http://www.omg.org/issues/uml2-rtf.open.html#Issue8761
I'm not sure how the OMG operates or if these will be resolved in the
UML 2.2 as there are no comments associated with them. If anyone has any
anything to add on this or has success using sequence diagrams, I'd
really be interested.
Regards,
Andrew.
Ar 23/07/2007 16:40, Scríobh Andrew Carton:
> Hi James,
>
> Thanks for answering. I understand that sequence diagrams are not
> suitable to model the entire behaviour of a system, but I only hope to
> generate code with the capabilities that they do offer. Any assistance
> you could give on this I would greatly appreciate. I think you mentioned
> in a previous post that there is a workaround by attaching expressions
> to messages - and I am thinking of doing this - if there is no other way.
>
> Thanks,
> Andrew.
>
> James Bruck wrote:
>> Hi Andrew,
>>
>> I'll work on a more detailed answer to this question but as a quick
>> note I
>> wanted to mention that if you are hoping to do code generation from
>> sequence
>> diagrams solely, I don't think that you will be successfull because of
>> issues with sequence diagrams ( I would encourage others to
>> aggree/disagree
>> with me in this newsgroup ).
>>
>> I think that ultimately you will be forced to use Activity diagrams to
>> fully
>> express the details required to generate code. You can of course use a
>> combination of Sequence/Activity to express details for code
>> generation (in
>> fact, I would think that Class/Structure/Activity/Statechart/Sequence
>> would
>> all be required since each describe different aspects of code
>> Structure/Behavior).
>>
>> It's been my experience that sequence diagrams are great for sketching
>> but
>> sometimes lack ability to express minute details.
>>
>> In this case, I don't think it's the UML API that is holding us back but
>> rather an interpretation of the spec. If we can figure this out from
>> the
>> UML spec, implementing this with the UML API should be straight forward.
>>
>> As I mentioned above, I will provide a more detailed answer as soon as I
>> work out all the details. In the meantime, I would suggest considering
>> Activity diagrams for code generation.
>>
>> - James.
>>
>>
>> "Andrew Carton" <cartona@cs.tcd.ie> wrote in message
>> news:f7qg9i$qve$1@build.eclipse.org...
>>> I'm trying to understand how to form valid sequence diagrams for code
>>> generation.
>>>
>>> I read the previous message on messages with return value and local
>>> variables but I still don't quite understand it.
>>>
>>> My first question is, how to pass the below as arguments to a message.
>>> As I understand it you can only create a ValueSpecification (and its
>>> subclasses) as an argument to a message (with its signature as an
>>> operation).
>>>
>>> The specification says you can pass these as values to the message.
>>>
>>> i) attributes of the sending lifeline.
>>> ii) constants.
>>> iii) symbolic values (which are wildcard values representing any legal
>>> value).
>>> iv) explicit parameters of the enclosing Interaction.
>>> v) attributes of the class owning the Interaction.
>>>
>>> For the first case, what exactly is an attribute of a sending lifeline
>>> and how do you create and reference it as a ValueSpecification? How do I
>>> reference as an argument an explicit parameter or an attribute of a
>>> class owning the Interaction.
>>>
>>> For return values, a previous post said to create an
>>> InstanceSpecification that refers to a ValueSpecification (I am assuming
>>> an InstanceValue here). The InstanceValue can then reference the
>>> InstanceSpecification and you can set the name as the variablename and
>>> the type as the type of the local variable. However, an
>>> InstanceSpecification is not really local as it belongs to a package and
>>> not an enclosing interaction or even the behaviourClassifier that the
>>> interaction belongs to. So this looks like it is a global instance?
>>> Also, how do you assign the return value to a class attribute that the
>>> lifeline represents?
>>>
>>> If someone has figured this out and can provide code I would really
>>> appreciate it as I am finding it quite difficult to understand how these
>>> are accomplished from both the specification and the uml2 api.
>>>
>>> Thanks,
>>> Andrew.
>>>
>>>
>>>
>>>
>>
>>
|
|
| |
| Re: Interaction Messages [message #623926 is a reply to message #474393] |
Fri, 20 July 2007 15:57 |
james bruck
Messages: 1724
Registered: July 2009 |
Senior Member |
|
|
Hi Andrew,
I'll work on a more detailed answer to this question but as a quick note I
wanted to mention that if you are hoping to do code generation from sequence
diagrams solely, I don't think that you will be successfull because of
issues with sequence diagrams ( I would encourage others to aggree/disagree
with me in this newsgroup ).
I think that ultimately you will be forced to use Activity diagrams to fully
express the details required to generate code. You can of course use a
combination of Sequence/Activity to express details for code generation (in
fact, I would think that Class/Structure/Activity/Statechart/Sequence would
all be required since each describe different aspects of code
Structure/Behavior).
It's been my experience that sequence diagrams are great for sketching but
sometimes lack ability to express minute details.
In this case, I don't think it's the UML API that is holding us back but
rather an interpretation of the spec. If we can figure this out from the
UML spec, implementing this with the UML API should be straight forward.
As I mentioned above, I will provide a more detailed answer as soon as I
work out all the details. In the meantime, I would suggest considering
Activity diagrams for code generation.
- James.
"Andrew Carton" <cartona@cs.tcd.ie> wrote in message
news:f7qg9i$qve$1@build.eclipse.org...
> I'm trying to understand how to form valid sequence diagrams for code
> generation.
>
> I read the previous message on messages with return value and local
> variables but I still don't quite understand it.
>
> My first question is, how to pass the below as arguments to a message.
> As I understand it you can only create a ValueSpecification (and its
> subclasses) as an argument to a message (with its signature as an
> operation).
>
> The specification says you can pass these as values to the message.
>
> i) attributes of the sending lifeline.
> ii) constants.
> iii) symbolic values (which are wildcard values representing any legal
> value).
> iv) explicit parameters of the enclosing Interaction.
> v) attributes of the class owning the Interaction.
>
> For the first case, what exactly is an attribute of a sending lifeline
> and how do you create and reference it as a ValueSpecification? How do I
> reference as an argument an explicit parameter or an attribute of a
> class owning the Interaction.
>
> For return values, a previous post said to create an
> InstanceSpecification that refers to a ValueSpecification (I am assuming
> an InstanceValue here). The InstanceValue can then reference the
> InstanceSpecification and you can set the name as the variablename and
> the type as the type of the local variable. However, an
> InstanceSpecification is not really local as it belongs to a package and
> not an enclosing interaction or even the behaviourClassifier that the
> interaction belongs to. So this looks like it is a global instance?
> Also, how do you assign the return value to a class attribute that the
> lifeline represents?
>
> If someone has figured this out and can provide code I would really
> appreciate it as I am finding it quite difficult to understand how these
> are accomplished from both the specification and the uml2 api.
>
> Thanks,
> Andrew.
>
>
>
>
|
|
|
| Re: Interaction Messages [message #624043 is a reply to message #474395] |
Mon, 23 July 2007 15:40 |
Andrew Carton
Messages: 104
Registered: July 2009 |
Senior Member |
|
|
Hi James,
Thanks for answering. I understand that sequence diagrams are not
suitable to model the entire behaviour of a system, but I only hope to
generate code with the capabilities that they do offer. Any assistance
you could give on this I would greatly appreciate. I think you mentioned
in a previous post that there is a workaround by attaching expressions
to messages - and I am thinking of doing this - if there is no other way.
Thanks,
Andrew.
James Bruck wrote:
> Hi Andrew,
>
> I'll work on a more detailed answer to this question but as a quick note I
> wanted to mention that if you are hoping to do code generation from sequence
> diagrams solely, I don't think that you will be successfull because of
> issues with sequence diagrams ( I would encourage others to aggree/disagree
> with me in this newsgroup ).
>
> I think that ultimately you will be forced to use Activity diagrams to fully
> express the details required to generate code. You can of course use a
> combination of Sequence/Activity to express details for code generation (in
> fact, I would think that Class/Structure/Activity/Statechart/Sequence would
> all be required since each describe different aspects of code
> Structure/Behavior).
>
> It's been my experience that sequence diagrams are great for sketching but
> sometimes lack ability to express minute details.
>
> In this case, I don't think it's the UML API that is holding us back but
> rather an interpretation of the spec. If we can figure this out from the
> UML spec, implementing this with the UML API should be straight forward.
>
> As I mentioned above, I will provide a more detailed answer as soon as I
> work out all the details. In the meantime, I would suggest considering
> Activity diagrams for code generation.
>
> - James.
>
>
> "Andrew Carton" <cartona@cs.tcd.ie> wrote in message
> news:f7qg9i$qve$1@build.eclipse.org...
>> I'm trying to understand how to form valid sequence diagrams for code
>> generation.
>>
>> I read the previous message on messages with return value and local
>> variables but I still don't quite understand it.
>>
>> My first question is, how to pass the below as arguments to a message.
>> As I understand it you can only create a ValueSpecification (and its
>> subclasses) as an argument to a message (with its signature as an
>> operation).
>>
>> The specification says you can pass these as values to the message.
>>
>> i) attributes of the sending lifeline.
>> ii) constants.
>> iii) symbolic values (which are wildcard values representing any legal
>> value).
>> iv) explicit parameters of the enclosing Interaction.
>> v) attributes of the class owning the Interaction.
>>
>> For the first case, what exactly is an attribute of a sending lifeline
>> and how do you create and reference it as a ValueSpecification? How do I
>> reference as an argument an explicit parameter or an attribute of a
>> class owning the Interaction.
>>
>> For return values, a previous post said to create an
>> InstanceSpecification that refers to a ValueSpecification (I am assuming
>> an InstanceValue here). The InstanceValue can then reference the
>> InstanceSpecification and you can set the name as the variablename and
>> the type as the type of the local variable. However, an
>> InstanceSpecification is not really local as it belongs to a package and
>> not an enclosing interaction or even the behaviourClassifier that the
>> interaction belongs to. So this looks like it is a global instance?
>> Also, how do you assign the return value to a class attribute that the
>> lifeline represents?
>>
>> If someone has figured this out and can provide code I would really
>> appreciate it as I am finding it quite difficult to understand how these
>> are accomplished from both the specification and the uml2 api.
>>
>> Thanks,
>> Andrew.
>>
>>
>>
>>
>
>
|
|
|
| Re: Interaction Messages [message #624045 is a reply to message #474507] |
Tue, 24 July 2007 11:00 |
Andrew Carton
Messages: 104
Registered: July 2009 |
Senior Member |
|
|
Hey,
If anyone else is looking at sequence diagrams, I have found some
unresolved issues at the OMG's UML2 RTF site relating to storing and
sending arguments with messages. The ones of particular interest:
http://www.omg.org/issues/uml2-rtf.open.html#Issue8785
http://www.omg.org/issues/uml2-rtf.open.html#Issue8786
http://www.omg.org/issues/uml2-rtf.open.html#Issue8787
http://www.omg.org/issues/uml2-rtf.open.html#Issue8788
http://www.omg.org/issues/uml2-rtf.open.html#Issue8899
http://www.omg.org/issues/uml2-rtf.open.html#Issue10591
http://www.omg.org/issues/uml2-rtf.open.html#Issue8761
I'm not sure how the OMG operates or if these will be resolved in the
UML 2.2 as there are no comments associated with them. If anyone has any
anything to add on this or has success using sequence diagrams, I'd
really be interested.
Regards,
Andrew.
Ar 23/07/2007 16:40, Scríobh Andrew Carton:
> Hi James,
>
> Thanks for answering. I understand that sequence diagrams are not
> suitable to model the entire behaviour of a system, but I only hope to
> generate code with the capabilities that they do offer. Any assistance
> you could give on this I would greatly appreciate. I think you mentioned
> in a previous post that there is a workaround by attaching expressions
> to messages - and I am thinking of doing this - if there is no other way.
>
> Thanks,
> Andrew.
>
> James Bruck wrote:
>> Hi Andrew,
>>
>> I'll work on a more detailed answer to this question but as a quick
>> note I
>> wanted to mention that if you are hoping to do code generation from
>> sequence
>> diagrams solely, I don't think that you will be successfull because of
>> issues with sequence diagrams ( I would encourage others to
>> aggree/disagree
>> with me in this newsgroup ).
>>
>> I think that ultimately you will be forced to use Activity diagrams to
>> fully
>> express the details required to generate code. You can of course use a
>> combination of Sequence/Activity to express details for code
>> generation (in
>> fact, I would think that Class/Structure/Activity/Statechart/Sequence
>> would
>> all be required since each describe different aspects of code
>> Structure/Behavior).
>>
>> It's been my experience that sequence diagrams are great for sketching
>> but
>> sometimes lack ability to express minute details.
>>
>> In this case, I don't think it's the UML API that is holding us back but
>> rather an interpretation of the spec. If we can figure this out from
>> the
>> UML spec, implementing this with the UML API should be straight forward.
>>
>> As I mentioned above, I will provide a more detailed answer as soon as I
>> work out all the details. In the meantime, I would suggest considering
>> Activity diagrams for code generation.
>>
>> - James.
>>
>>
>> "Andrew Carton" <cartona@cs.tcd.ie> wrote in message
>> news:f7qg9i$qve$1@build.eclipse.org...
>>> I'm trying to understand how to form valid sequence diagrams for code
>>> generation.
>>>
>>> I read the previous message on messages with return value and local
>>> variables but I still don't quite understand it.
>>>
>>> My first question is, how to pass the below as arguments to a message.
>>> As I understand it you can only create a ValueSpecification (and its
>>> subclasses) as an argument to a message (with its signature as an
>>> operation).
>>>
>>> The specification says you can pass these as values to the message.
>>>
>>> i) attributes of the sending lifeline.
>>> ii) constants.
>>> iii) symbolic values (which are wildcard values representing any legal
>>> value).
>>> iv) explicit parameters of the enclosing Interaction.
>>> v) attributes of the class owning the Interaction.
>>>
>>> For the first case, what exactly is an attribute of a sending lifeline
>>> and how do you create and reference it as a ValueSpecification? How do I
>>> reference as an argument an explicit parameter or an attribute of a
>>> class owning the Interaction.
>>>
>>> For return values, a previous post said to create an
>>> InstanceSpecification that refers to a ValueSpecification (I am assuming
>>> an InstanceValue here). The InstanceValue can then reference the
>>> InstanceSpecification and you can set the name as the variablename and
>>> the type as the type of the local variable. However, an
>>> InstanceSpecification is not really local as it belongs to a package and
>>> not an enclosing interaction or even the behaviourClassifier that the
>>> interaction belongs to. So this looks like it is a global instance?
>>> Also, how do you assign the return value to a class attribute that the
>>> lifeline represents?
>>>
>>> If someone has figured this out and can provide code I would really
>>> appreciate it as I am finding it quite difficult to understand how these
>>> are accomplished from both the specification and the uml2 api.
>>>
>>> Thanks,
>>> Andrew.
>>>
>>>
>>>
>>>
>>
>>
|
|
| |
Goto Forum:
Current Time: Fri Apr 19 23:40:55 GMT 2024
Powered by FUDForum. Page generated in 0.03357 seconds |
] 
Search
Help
Register
Login
Home
