Re: Is garbage collection here yet?

Neil Madden <nem@cs.nott.ac.uk> wrote:
> Andreas Leitgeb wrote:
>> My definition: garbage is, what the programm cannot access anymore.
>> Your (appearant) definition: anything the program doesn't need anymore.

> It is generally not possible for an object to become unreachable
> in Tcl in the same way that one can become unreachable in Java.
> So your definition would seem to rule out GC for objects in Tcl.
Yes, indeed.
GC would only become important, if we had objects that are not
tied into the usual places (cmd,var,nsp). I don't want to go
that route, but that alone doesn't mean it might not happen some
day...

>> References are "not Tcl" in about the same way as dynamic code
>> or accessing caller's local variables is "not Java".
> You'd have to define what you mean by reference here. Certainly
> Tcl has means to reference things -- variables etc.

I thought it was clear from the context, that "References" in
Tcl *would* be like those of jim or hecl: somewhat like anonymous
variables.

Anonymous variables are (until someone convinces me of the opposite)
a contradiction to the "everything is a string": as soon as a
reference can be stringified(and later restored), it can no
longer be determined at any point in time whether any reference
exists to a given object or not, thus thwarting GC.
Received on Sun Apr 30 02:35:16 2006