[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

(FC-Devel) Further musings on direction



Jacques Bergeron wrote:

> Can I describe your approach like this:
> 
> - CORBA server (ORB);
> - UNIX based;

Not entirely UNIX based.  MICO (the ORB I'm using) apparently compiles
under MS Visual C++ and I have compiled it under cygwin on NT.
 
> - C++ as the programming language;
> - On-line model sharing between the clients i.e. if one client change a part
> of the model then all other clients will be notified immediately.

This is one option.  Other options include:
1)  Lock packages before update is allowed.
2)  Use pserver CVS on the XMI output (using a local repository).

Don't know how well these fit into the ModelManagement bit because I
haven't looked at that part in depth yet.  I'm guessing that there'd
have to be a separate Freecase module to deal with this kind of thing. 

> If so then a client modeling tool would have to change its model management
> component to integrate into the repository (more work but not impossible in
> the case of Argo anyway).
> 
> I have no prior experience with CORBA

Neither did I before I started.  Only an interest.

> and C++ is a distant memory (had to do
> management work for some years).

I code mostly in Ada by day.  This is my first C++ project, although
I've done tons of C.  I've done Shaler Mellor OOA but only read about
UML.  Add to this the research I've been doing into XML (for the XMI
part), SVG (for the diagrams) and DOM (when I had a momentary wobble
about the way I was approaching the problem) and you can see why
progress has been slooooowwwww.

I'm having a great time though!

> Working on a JAVA IIOP client interface
> would be interesting though espacially since the CORBA IDL physical
> interface to the UML model is clearly defined in the OMG specs. The Java /
> IIOP mapping also seems to be clearly defined by Sun. If thing go well that
> could lead to an on-line "bridge" between the repository and Java modelling
> tools (Argo and others).
> 
> Could that be of some interest to you?

I'm mono-maniacal about the repositiory since this seems to be the heart
of any CASE tool.  If you want to write a new Java client or modify Argo
to use Freecase then go for it if that is what interests you.  I'll try
to minimise the number of NO_IMPLEMENT exceptions you get. :o)

> On another subject would your repository be directly reachable (on-line)
> only by a LAN or also by the Internet. On the latter case you may walk into
> performance and access problems (ex. firewalls) that I would like to discuss
> with you.

I have suggested before that:

"For my home machine (dial up access to the net) I'd run the CORBA
server on the
same machine as the client and have the backend (for example cvs) run on
a
remote machine.

While at work, I'd expect the company to have a central super-server
which
exposes the CORBA server interface to the LAN."

Chris Moore

--
Sig pending!