Re: function to resize a UML class object's extent?

[Cyrille Chepelov <chepelov calixo net>]

Le ven, nov 30, 2001, à 07:55:22 -0700, Andrew S . Halper a écrit:

> There are commuters in France? (!)

That must sound strange to you, eh ? Some people actually do about 1 hour 
each way. For me, that wasn't a too bad commute, at the beginning: 30 min. 
each way, but unfortunately, that was 30 min. of driving (the day there is a 
HUD, windshield-mounted html reader with eye movement recognition available in 
cars 10% more expensive than what I can afford, I guarantee you I'm buying)
But add 1.5 hours idling alone in the office (eating basically salads, döner
kebab or big macs), and it quickly grows tiresome. Soon, that's going to be
a matter of 5 min. walking. Phew.

> Would it be preferable for me to 1) write some fiendish electric Frankenstein of
> StdProp and the older interface in my plug-in, or 2) attempt to write the UML
> StdProp bits that I need?
>
> I'll admit that most of the StdProp C-as-OOL code I've seen is at a level of C
> technical proficiency that is beyond me, but on the other hand, I do an
> abundance spare time, and could probably make a decent try at it eventually.

What I envisioned was factorise as much code as possible; through StdProp or
through any other means (for instance, the improvement I suggested a couple
days ago wrt arrow type selection widgets proceeds from that school of
thought. For me, less code for about the same result means less bugs).

Enough of philotheoretical ramblings; C-as-OOL is really less complicated
than what it looks like; one just has to explicitly drag around "this" and
build one's own vtables (OTOH, you can do some pretty Python-esque things in
C-as-OOL you can't in C++; like, alter at runtime the vtable of an instance
to change the implementation of a method. Sounds complicated ? It's not, if
you see a function pointer not as a function pointer, but as a single-entry
vtable...). So, to finish, my advice wrt what you want to write: it would
probably means less code and more reuse if you could write the
controls/property types you lack as more StdProp types; you'll certainly
want to investigate (and debug <grin/>) the "action" field I wrote support
for (in the view of stdpropising the UML Class), but never got around to use
more than in a "hello, world" example (you'll find examples of it if you dig
the ChangeLog, circa end of July 2001).

In any case, good luck. I'll probably remain a spectator for a while.

        -- Cyrille

-- 
Grumpf.