[Uml-user] basic questions

[Bryan Lawrence]

Hi Jonathan

Thanks for getting back to me.

On thinking about it, it does make sense to try and remain language agnostic 
in the UML ... after all, it's not supposed to be a RAD tool per se ... 
however, I do think any UML tool should support an array datatype because 
that's a concept that is universal in real programming languages (ok, so I 
confess to being a fortran programmer at heart).

I'm not quite sure I understood your last sentence. Was that bit the bit in 
your reply addressing my question about the attributes which are forced to be 
integers (and nothing else - unlike the return values from operations, which 
are allowed to be any builtin datatype). Did you mean then, that it shoudn't 
do anything at all, or it should behave the same as the operations?

Cheers,
Bryan

On Saturday 27 March 2004 17:40, Jonathan Riddell wrote:
> On Wed, Mar 24, 2004 at 07:17:01PM +0000, Bryan Lawrence wrote:
> > 1) If I create a class, why is it that I can only create attributes which
> > are integers via the pull-down menu?   (I can type in my own datatypes,
> > but then they turn up as UML datatype objects - I would have thought I
> > didn't need to do that for language builtins like booleans etc - which
> > indeed do turn up as options for return values from operations that I
> > create for the class).
>
> booleans and other builtin types are datatypes.
>
> It's because I havn't added datatypes for Python yet.  The question is
> should we have a standard set of datatypes for all languages and
> translate them as code is exported (since UML is supposed to be
> language indepenant) or have specific datatypes for each language.
> The first seems more correct.
>
> > 2) I think the above is language independent. Python-wise, is there any
> > chance I can get python types like dictionary and lists as datatypes
> > without having to create them myself?
>
> Actually I don't know if inbuild data structures like dictionarys,
> lists and (in C type languages) arrays should be datatypes, I suspect not.
>
> There seems to be a problem, when you add an attribute of type
> new_class1 or whatever it tries to create a datatype for you, which it
> shouldn't.
>
> Jonathan