R devel - Jul 2009 - destructor for S4 class objects in analogy to C++

Hi all,

I'm wondering if there is a way to define a destructor function (to free
memory) for S4 class objects in analogy to C++?
rm() combined with gc() does not seem to be a good idea (Chambers, 2008).
So could it be done on the C/C++ level or is it even already available in
the "internals"?

Many thanks,
David

	[[alternative HTML version deleted]]

if you are using external pointers, then R allows you to set a
finalizer function to be called on the pointer.

http://cran.r-project.org/doc/manuals/R-exts.html#External-pointers-and-weak-references

see examples in RODBC package.

and here too:
see the "connFinalizer" function
http://github.com/armstrtw/unifieddbi/blob/3b21da83f96de231fb14a1adaf9150fa1bc6a277/src/interface.cpp

-Whit



On Tue, Jul 21, 2009 at 1:00 PM, David
Scherrer<dasmailinglists at gmail.com> wrote:
> Hi all,
>
> I'm wondering if there is a way to define a destructor function (to
free
> memory) for S4 class objects in analogy to C++?
> rm() combined with gc() does not seem to be a good idea (Chambers, 2008).
> So could it be done on the C/C++ level or is it even already available in
> the "internals"?
>
> Many thanks,
> David
>
> ? ? ? ?[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

David,

On Jul 21, 2009, at 13:00 , David Scherrer wrote:
> I'm wondering if there is a way to define a destructor function (to  
> free memory) for S4 class objects in analogy to C++?
The analogy doesn't work, because in C++ you have pass-by-reference  
semantics, but in R you have pass-by-value. Hence in R there is no  
"one" object that you could finalize - there may be dozens of copies  
of your S4 object, so which one would you finalize? However, you can  
register a finalizer for reference-semantics objects (if that's what  
you really have), see ?reg.finalizer (and on C level  
R_Register*Finalizer* functions).

Cheers,
Simon

> rm() combined with gc() does not seem to be a good idea (Chambers,  
> 2008).
> So could it be done on the C/C++ level or is it even already  
> available in
> the "internals"?
>
> Many thanks,
> David
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
>