On 11/01/2007, at 3:44 AM, Charles Yeomans wrote:
On Jan 10, 2007, at 2:29 PM, Stefan wrote:
The RB framework
adds a significant amount of additional CPU cycles by mapping RB's
lib to native libs.
Do they? I think most of that mapping is resolved during
compilation and linking.
It is theoretically possible for a cross-platform C++ framework to
resolve most of its mapping at compile time. My own PP2MFC does
exactly that which is why it is so efficient, and why it had to be
written in the first place as none of the others available in 1997
did so. So, it serves as an existence proof.
PP2MFC accomplishes this by having the compile-time mapping occur
within your own application, using c++ inline functions to remap APIs.
I would be very surprised, therefore, if the RB compiler/linker can
optimise away a lot of the mapping as the compiler seems very light
on optimisation (based on public statements) and moves in RB200x have
been to make more things virtual.
At the very least, things which may have previously been plain
function calls to the OS, and may even have been inlines, are now
hidden behind a virtual function.
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
