Re: circular unit dependency

["John G. Volan" <johnvolan@sprintmail.com>]

Jon S Anthony wrote:
> 
> In article <3386d96f.171920@noticias.ibernet.es> jdlopez@caminos.recol.es writes:
> 
> >      I am making a program with 2 classes that each one depends on the
> > other:
...
> Uh-Oh.  The recursive types across package boundaries issue again!
> Run for it!!  :-)
> 
> This is an old issue that (rightly, I suppose) keeps coming up now and
> again.  John Volan has a kind of generally agreed upon idiom for
> getting around this issue using generics.  Maybe he will post his old
> notes on the reasons he got where he did.

Well, this time instead of posting the whole darn FAQ on comp.lang.ada,
I thought I'd take a couple of days, polish it up, turn it into HTML,
and publish it as a website, something I've been meaning to do for a
long time...it looks real spiffy in my browser now...but...I have just
discovered that my ISP doesn't provide personal web space!! (Huh?? I'm
sure I asked about that when I signed up! Arggh!) Oh well, I know
someone who might set me up with a quickie website cheap...will let
everyone know shortly...

> Or maybe it should be put out on www.adahome.com??

I'd love it if it were on adahome or someplace comparable.  Ages ago I
did mail my whole FAQ to Magnus Kempe in hopes he would install it into
adahome somewhere, or maybe the Ada Programmer's FAQ...but he refused.
He said it was because I stuck the LGPL on the generic Identity package.
(By the way, I've dropped the whole "object class identity" discussion
as being too abstruse, so I've changed the name of my generic package to
"Forward", signifying "forward declarations.")

Magnus is apparently allergic to the LGPL (or maybe he's just allergic
to my little generic gizmo, I dunno... :-) ).  At the risk of
resurrecting another monster thread, anyone have any advice on what
license verbiage I should use instead? I don't want to place any
restrictions on the use of my Forward package (i.e., I want to let
proprietary commercial stuff incorporate it without catching the free
software virus). But on the other hand, I want to keep copyright control
over it to prevent anyone from turning it proprietary and keeping others
from using it.  I thought the LGPL was good enough for that but now it
seems I'm wrong...

> Or you can read it in the Ada=>IDL mapping
> document (where we used it to get around the problem of mapping
> recursive IDL specs...)

Yeah, you might be able to puzzle out the generic forward declaration
pattern by looking at the CORBA.Forward package (since it directly
incorporated my idea). But unless you're actually doing distributed
programming with CORBA tools, you're not going to be able to use that
directly.  Better to look at my FAQ...as soon as I can put it
somewhere... :-)

> BTW, whatever happened to Tucker's plan for the fix for this?  The
> "with P.T" solution?  I thought that was going to be an ACE thing or
> some such, but haven't heard anything about it since.

Yeah, whatever did happen to Tucker's "with type" proposal?  I remember
he said this was something that really did need to be addressed "sooner
rather than later" -- i.e., before AdaOY.

(By the way, I've added a description of Tucker's "with type" notion
into my FAQ.)

-----------------------------------------------------------------------
Internet.Usenet.Put_Signature 
  (Name => "John G. Volan",  Home_Email => "johnvolan@sprintmail.com",
   Slogan => "Ada95: The World's *FIRST* International-Standard OOPL",
   Disclaimer => "These opinions were never defined, so using them " & 
     "would be erroneous...or is that just nondeterministic now? :-) ");
------------------------------------------------------------------------