Re: obvious extension of o-xml to all forms of programming, and need for import facility from existing code

Hi,

I sent this message a couple of days ago and it didn't seem to make it to
the list even though I authenticated my message when asked to do so by the
automatic spam manager.  Anyhow, here it is again.

One thing that could be worth thinking about when talking about bringing in
functional specs is literate modelling (Jim Arlow). It would be nice if
textual/diagrammatic artifacts like literate modelling documents (outputted
in XMI ?) could serve as a basis for functional spec definition, which could
then be turned into o:XML code and to user manuals by XSLT (?) and finally
used as a validating artifact for test cases.  XP user stories could be
embedded into literate modelling documents and therefore be linked more
closely to code than if they were mere text documents held separately.

I guess there could be a way of specifying the granularity at which one
wants to "manage" one's code (not every helper function or class would show
in UML diagrams ...) so that only some of the o:XML code could be
specifically tied to textual/diagrammatic artifacts such as literate
modelling documents.  Maybe a tool like xMLP could fit in well with such a
scheme ?

Yves Moisan

"Jeff Davies" <jeff-YhAa5rnyZPVjitOJ6T5g7Lyq0pUa53/[email protected]> a
écrit dans le message de
news:200303172120.24892.jeff-YhAa5rnyZPVjitOJ6T5g7JU/[email protected]
o-xml is obviously transformable (xslt or special sax program) to bytecode,
sourcecode in c/java etc or even direct to assembly language as an output
medium.

Of course, there is a large requirement for importing existing code to
o-xml.

Perhaps a diagram somewhere showing:

C++
Java
C -->  import utility -->  o-xml --->xslt etc -> Java src /x86 code etc

for those who are new to the concepts and need to be sold o-xml would be a
good thing on the website.

One of the great things I think you can do with code in XML format is
enforce
rules such that a functional spec for a function (for extreme programming
this would be called 'the story') is written first, followed by unit test
cases (perhaps even parameterised ... like    20e4>x>4). Xlinks to test
results could also be stored in the same document  (assume a versioned XML
storage system here).
Similarily, code auditing can be enforced.

The advantages to having a functional spec (with audit points in it) would
be
that you could transform the program and obtain the functional spec as a
PDF.
As well as enforcing the functional spec is updated by the same guy who is
changing the code, at the same time as he's coding.

Commercially, I think a tie up with Rational Rose (now owned by Borland)
would
be worth contemplating. Dancing to close to Microsoft might be a good way to
get ripped off according to the horror stories I've heard (although they
might not be true, I simply don't know for a fact).

Jeff Davies