Object Migration -- Not Part of CORBA -- WHY ?

vadlamv@ibis.dun.nielsen.com (Vish Vadlamani)
Mon, 5 Dec 1994 15:59:10 GMT

          From comp.compilers

Related articles
Object Migration -- Not Part of CORBA -- WHY ? vadlamv@ibis.dun.nielsen.com (1994-12-05)
| List of all articles for this month |
Newsgroups: comp.object,comp.lang.c++,comp.compilers
From: vadlamv@ibis.dun.nielsen.com (Vish Vadlamani)
Keywords: question, C++, OOP
Organization: Univ. of Florida CIS Dept.
Date: Mon, 5 Dec 1994 15:59:10 GMT

Memory Resident Library -- In the context of Object Migration


In the context of Object Migration from a source to destination machines I
had assumed a C++ and binary compatible OS on both ends. This was done to
limit the scope of discussion to actual implementation details rather than
as an exercise in theoretical dialogue.




The concept proposed was that the compiler at compile time produce a piece
of code which could be loaded into memory. This is not linked at link time
or run time but loaded into memory at specific request from the
application itself.


for example:
The first instance of the object could possibly be a good candidate for
loading its methods into memory. ( I see a problem already -- What then
of the class methods ?? )




Then members of this Memory Resident Library can actually gain control
when they are invoked as instance methods.


( Again a lot of eyebrows will be raised for this unconditional
transfer of control )






Are there any compiler gurus who can share their thoughts on whether
the above is a feasible option. Or if such a thing already exists
in some other language implementation.




Vish Vadlamani




--
Nielsen Media Research
375 Patricia Avenue
Dunedin
FL 34698
Voice: (813)738-3000 ext 2659
FAX:(813)738-3007
--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.