char [][char[]] a;
like above that would result a map between char[] and char[], there must be some memory operations. so would it call object.d's new operator's implementation?
and char []a;
a.length=34; <--would that call object.d's new implementation?

I wanna use full d's feature to write a kernel. so just ask whether d is suitable to write a kernel.

-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/

davidl wrote:
> char [][char[]] a;
> like above that would result a map between char[] and char[], there must  be some memory operations. so would it call object.d's new operator's  implementation?
> and char []a;
> a.length=34; <--would that call object.d's new implementation?

I don't know much about kernels or the insides of D's memory management, so I can't help there, sorry.

> 
> I wanna use full d's feature to write a kernel. so just ask whether d is  suitable to write a kernel.
> 

What I can do is point you to people who are also writing a kernel in D.  Check out Titan on dsource.
http://www.dsource.org/projects/titan
http://www.dsource.org/forums/viewforum.php?f=73

btw, you'd probably be better off posting this on digitalmars.D.learn or maybe digitalmars.D.  This stuff doesn't seem related to GDC and would be handled better in those other branches of the newsgroup.

I have tested GDC's implementation. I use gdc to compile a file containning that char[][char[]] a;
I found it would use the internal file GC to do memory operation, and _d_new to new objects, I also
found it use d_invariant??? few more characters in that name, don't know what that is. hope some one
could tell me. But I think if we properly inplemented the internal GC functions, we can write a kernel
in D.





-- 
使用 Opera 革命性的电子邮件客户程序: http://www.opera.com/mail/