www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Unfortunately delete is a keyword

reply Rory Mcguire <rjmcguire gm_no_ail.com> writes:
Inspired by the recent thread "Proposal for dual memory management" I 
wondered how long it would take to change scoped into heaped.
Now I wish I could call the free method delete.
I didn't put the full test source here because I had to copy some stuff from 
phobos svn.

Please let me know what you think, is this already done?

-Rory

code:
 system auto heaped(alias T,Args...)(Args args) {
    struct HeapAlloc {
        private void *buf;
        T cpayload() {
            return cast(T)(cast(byte[]*)buf);
        }
        alias cpayload this;

        void free() {
            destroy(cpayload);// not in 2.046
            // not sure what this does!!!
            if ((cast(void**) cpayload)[1]) {
                _d_monitordelete(cpayload, true);
            }
            .free(buf);
        }
    }
    void* mem = malloc(__traits(classInstanceSize, T));
    if (!mem)
        throw new Exception("no mem");

    emplace!T(cast(void[])mem[0..__traits(classInstanceSize,T)],args);

    HeapAlloc ret;
    ret.buf = mem;
    return ret;
}

Usage:
    auto a = heaped!A;
    auto b = heaped!A(10);
    
    assert(a.i==3);
    assert(a.getI()==3);

    assert(b.i==10);
    assert(b.getI()==10);
    
    a.free();
    b.free();
Jul 29 2010
parent reply Kagamin <spam here.lot> writes:
Rory Mcguire Wrote:


 Inspired by the recent thread "Proposal for dual memory management" I 
 wondered how long it would take to change scoped into heaped.


You can also try to implement switch, for and while statements in a similar
way. That's a whole bunch of keywords.
Jul 29 2010
next sibling parent reply =?UTF-8?B?IkrDqXLDtG1lIE0uIEJlcmdlciI=?= <jeberger free.fr> writes:
Kagamin wrote:

 Rory Mcguire Wrote:
=20

 Inspired by the recent thread "Proposal for dual memory management" I =





 wondered how long it would take to change scoped into heaped.


=20
 You can also try to implement switch, for and while statements in a sim=


ilar way. That's a whole bunch of keywords.

	Then you get scheme (or felix) where almost all the language
constructs are actually part of the std library...

		Jerome
--=20
mailto:jeberger free.fr
http://jeberger.free.fr
Jabber: jeberger jabber.fr
Jul 29 2010
parent Kagamin <spam here.lot> writes:
Jérôme M. Berger Wrote:


 Then you get scheme (or felix) where almost all the language
 constructs are actually part of the std library...


every language designer's dream :3
Jul 29 2010
prev sibling parent Rory Mcguire <rjmcguire gm_no_ail.com> writes:
Kagamin wrote:


 Rory Mcguire Wrote:
 

 Inspired by the recent thread "Proposal for dual memory management" I
 wondered how long it would take to change scoped into heaped.


 
 You can also try to implement switch, for and while statements in a
 similar way. That's a whole bunch of keywords.


hmm, there was a benchmark here somewhere where somone (bearofile?) compared
if and if else performance to a switch statements performance and if did get
better performance for some things.

:D
Jul 29 2010