Re: ObjectiveC accessors

[Marcel Weiher]

On 1 Jun 2004, at 11:00, Helge Hess wrote:

> On Jun 1, 2004, at 10:14, Marcel Weiher wrote:
> > > > > I agree that this is not necessarily a good idea ;-), but notably 
> > > > > Boehm-GC can be used as a drop-in replacement for the malloc 
> > > > > library (without altering semantics!)
> ...
> > If you use 2 different mallocs, then it's obviously not a drop in 
> > replacement.
>
> Hu? Boehm-GC *is* a drop-in replacement

The malloc is a drop-in, as long as you don't turn scanning on (i.e. 
not using the GC).  As soon as you turn the scanning/GC on, it isn't 
semantically compatible any longer.

>  (nothing else was stated), using it as such allows you to more 
> reliably use the conservative GC functionality *in addition*.



> > Sure.  It just isn't semantically compatible, and cannot be.
>
> It absolutely is.

NO.

>  If you use the GC malloc instead of system malloc, it isn't

PRECISELY.

> (I can't remember that someone ever asked for that?), if you use it as 
> a drop-in replacement, it is perfectly malloc compatible

And isn't GC.  Duh.

>  and provides additional functionality on top of that.
>
> You are probably confused by the name of the software "BoehmGC"

Not at all.

>  - it actually does not only provide the GC,

Yes, but we were talking about GC.

>  but also a library that can be used instead of system malloc.

But using only that isn't GC.  And we were talking about GC.

>  Nothing else was said, see my initial statement on the top of the 
> mail.
> If you read that as "GC is available at C system level with C 
> compatible behaviour", you read it wrong

No, I didn't read it wrong.  We were talking about adding GC.

>  (this is obviously nonsense ;-)

Precisely.

>  Maybe we can agree on that and stop the thread ...

:-)

Marcel