Sequence Diagramming

Thread view:

Iain - 13 Feb 2006 18:47 GMT

Is there such a thing as a programme that will let me create, let us
say, ten sequence diagrams, and forge from that a single normalised
template of source that facilitates all those courses of action?

~Iain

Reply to this Message

Oliver Wong - 13 Feb 2006 18:53 GMT

> Is there such a thing as a programme that will let me create, let us
> say, ten sequence diagrams, and forge from that a single normalised
> template of source that facilitates all those courses of action?
>
> ~Iain

If you use Microsoft Visio, it can do this, but on the condition that
you also provide Class diagrams.

In fact, I believe Visio ignores any semantic information you provide in
the Sequence diagrams and generates it templates entirely from the
information in the Class diagrams.

- Oliver

Reply to this Message

Iain - 13 Feb 2006 19:17 GMT

> > Is there such a thing as a programme that will let me create, let us
> > say, ten sequence diagrams, and forge from that a single normalised
[quoted text clipped - 8 lines]
> the Sequence diagrams and generates it templates entirely from the
> information in the Class diagrams.

It seems to me as if nobody sees it as useful to take a sequence
diagram as input.

Don't you think it would be useful to be able to list the main possible
threads of action and have a template code drawn from that? You could
exapt later it for other uses if need be.

~Iain

Reply to this Message

Oliver Wong - 14 Feb 2006 18:17 GMT

>> > Is there such a thing as a programme that will let me create, let us
>> > say, ten sequence diagrams, and forge from that a single normalised
[quoted text clipped - 16 lines]
> threads of action and have a template code drawn from that? You could
> exapt later it for other uses if need be.

I'm not a UML guru, but I have the feeling that "what is going on" in a
sequence diagram is always clear. For example, when you have an arrow going
from one lifebar to another, what does that mean? Method invocation? TCP/IP
Socket communication? XML SOAP transmission? Inter-thread communication?

What is the significance of the length of the lifebars? How do these two
diagrams differ in semantics?

<diagram A>
[Object 1] [Object 2]
| |
|-| |
| | |-|
| |------->| |
|-| | |
| | |
| |-|
| |
</diagram A>

<diagram B>
[Object 1] [Object 2]
| |
|-| |
| | |-|
| |------->| |
| | | |
| | | |
| | |-|
|-| |
| |
</diagram B>

Furthermore, I believe sequence diagrams are not apt for representing
all the branches when a conditional construct (e.g. if statement) might
appear, or iteration constructos (while loops, for loops, etc.), all of
which are pretty central to the code that actually appears within the
methods.

If all we care about are the existences and signatures of the methods
themselves, a class diagram seems like a more ideal tool for this.

All a sequence diagram buys us is one possible ordering in which the
methods may be invoked, and the only thing we can use that in code
generation is to filling in the body of methods, but like I said, it's not
enough information to actually generate useful method implementations.

- Oliver

Reply to this Message

Richard Wheeldon - 13 Feb 2006 19:50 GMT

Rational Rose will allow you to create sequence diagrams based on
classes in a previously created class diagram. I'd imagine TogetherJ
will allow the same,

Richard

Reply to this Message

Iain - 13 Feb 2006 20:13 GMT

> > Is there such a thing as a programme that will let me create, let us
> > say, ten sequence diagrams, and forge from that a single normalised
[quoted text clipped - 3 lines]
> classes in a previously created class diagram. I'd imagine TogetherJ
> will allow the same,

I was more curious about a program that extracts a class structure from
a large number of already existing sequence diagrams.

~Iain

Reply to this Message

Sequence Diagramming

Free Magazines