Re: guile-gtk internals doc

[Kevin Ryde]

Andy Wingo <address@hidden> writes:
>
> would be reference leaks (notably, with attached signal handlers)

I suspect handlers are the only real problem, everything else being a
hierarchy.  Of course it's still a good thing if one gc pass can notice
an entire tree can be collected in one go, instead of one layer per gc.

> one day give it a try (again).

In the fullness of time I imagine, or at least I certainly don't propose
hugely expanding the guile-gtk side :-).  But I'm fairly browned off by
guile and scheme at the moment.  I seem to be the only one who gives the
proverbial rodents rectum about compatibility, the procession of bugs is
unending, and r6rs is a long way from what you really want in an
extension language.  Without the virtues of small/stable/simple then may
as well go to one of the big lisps, or even perl.  The scary thing is I
think perl used carefully is looking pretty attactive!