On Wed, 23 May 2001 17:44:03 -0400, Greg Lopp <[EMAIL PROTECTED]> wrote:
>Definitely interested. Always like a good challenge....
Wonderful!
>Can you provide more information about the scope of the problem you need
>solved? I understand that this new C++ compiler uses a different calling
>convention, but that won't cause any problems as long as C++ is
>calling/invoking C++. Where exactly will this not be the case? (two
>files in xpcom/reflect/xptcall/src/md/unix ?)
There is an FAQ that answers some of these questions at:
http://www.mozilla.org/scriptable/xptcall-faq.html
But to answer your question directly, my understanding is that the
xptcall code allows you to have an object written in C++, another
object written in java script, and a third object written in Python.
These objects can invoke each others methods with out having to care
which language the other object is written in. From C++ code they
all look like C++ objects. And when a java script object calls a C++
object with an overridden virtual method, it invokes the correct method.
To make this work, you have to be able to build call frames for methods on
C++ objects. This requires knowing the calling convention and how to find
the vtables. Hopefully the information that intel gave me is enough.
>Also, it looks like your bug report got cut off or something.
Its a cut and paste problem. I think all the information is there. The
problem is that intel has their own web based bug system which they are
tracking this issue with. I just did a cut-n-past from one mozilla window
to another which is why the formatting is wierd.
Thanks again,
Jim
