Le lun, nov 26, 2001, à 05:08:22 -0800, Eric Merritt a écrit:
> Neil,
>
> Thats not a bad idea. Speaking completly blindly and
> with no foundation in reality, labels would probably
> not be that bad to impliment.
That could be developed fairly easily (once one feels confident enough to
write a C Object, that is), by using the approach used by the flow label in
objects/SADT/annotation.[ch] (check out samples/SADT.dia for an instance of
these in action). Then it would just be a matter of adding a connection
point at the middle of the various UML relation objects, where that makes
sense (and that can be done in a completely upward-compatible way). [to
limit the user as little as possible, it may make sense to retrofit the
Connpoint_line approach used by SadtArrow into the UML relations, but that
may be overkill).
Happy hacking (2) !
-- Cyrille
--
Grumpf.