D vs VM-based platforms

BCS ao at pathlink.com
Tue May 1 12:32:30 PDT 2007


Reply to Benji,

[...]

Most of your rebuttal basically says that programs in VM can do X without 
the coder having to do something different than they would if they didn't 
do X. There is (lots of big problems aside) a simple solution to this problem 
in native code: Don't allow the coder to NOT do X, requiter that all classes 
be COM objects, always compile in debugging and profiling symbols, heck maybe 
even a full net-centric debugger. It seams to me (and I could be wrong) that 
the way that the VM languages get all of these advantages of these options 
is by not making them options, they are requirements.

The only case that all of that doesn't cover is sand boxing. Well, something 
is going to have to run in native code, so why not make native code safe? 
Allow a process to span off a thread that is native code but sand boxed: 
some OS API's don't work, it has a Read Only or no access to some part of 
ram that the rest of the process can access. In short If security is such 
a big deal, why is the VM doing it instead of the OS?





More information about the Digitalmars-d mailing list