COMPARATIVE STUDY: Pure OOPLs VS Hybrid OOPLs (long)

COMPARATIVE STUDY: Pure OOPLs VS Hybrid OOPLs (long)

[hakim]

Do you have something more to add or to publish about " comparative
 study : PURE OOPLs VS HYBRID OOPLs" ? please write it to me
 hakim@pl.jaring.my or fahr820@tmsk.itm.edu.my
 Fahrul Hakim
 Msc. Computer Science M.I.T SHAH ALAM SELANGOR, M'SIA.

 this is my article : please read it.

 PURE OOPLs Vs HYBRID OOPLs : Uniformity versus efficiency (comparative
 analysis)
 ---------------------------------------------------------


   in pure OOPLs , all oo techniques (in particular inheritance,
 polymorphism, and dynamic binding) can be applied to all sorts o
 data, regardless of whether they are elementary or not.  The relust is a
 uniform and consistent thougt model that serves as the basis for the
 construction of all algorithms.  This is very convenient for the
 programmer, as no mental switching between two kinds of operations is
 necessary.
   However, there is price to pay for such uniformity.  Dynamic
 binding
 is involved whenver an operation with a simple integer number is
 performed, thus leading to a considerable slowdown.  This is the
reason
 why hybrid languages are generally regarded as efficient, while pure

 OOPLs are deemed slow.
   The main reason why hybrid languages are more efficeint lies in the
 fact that they do not support polymorphism with non-objects.  This
has
 two important consequences:
 1) The compiler can statically determine the storage requirements for
 conventional data. Whereas an object variable can refer to objets of
 different sizes, a non-object variable will always contain data of a
 fixed size.  This means that a certain storage area can be reserved
for
 every conventional variable at complile time.  It is thus not
necessary
 to access components of such variables via pointers, whereas in pure
 oopls such indirect access cannot be avoided.

 2)Elementary data types can be optimized.  When a variable x is of
   the type integer , static typing reles ensure that x will always

   contain an integer number at run time.  The compiler can use the

   "knowledge" to treat the variable more efficeintly.  For example,

    special processor instructions can be used to optimize frequent

   operations such as the statement x:=x+1.

   There is no simple solution to this problem.  Compared to
   conventional data, additionla coses cannot be avoided whehn objects
are
   manipulated. In hybrid languages, these costs can be avoided simply
by
   avoding objects altogether, but that would be like emptying the
baby
   out with the bath.

 Several attempts have been implemented to avoid the run-time costs of
 operations with elementary objects. For e.g. sophisticated
compilation
 techniques (in particular, customized compilation, message  inlining,
message splitting, and type prediction) are used by the Self
compiler
to achieve high performace.  In omega, elementary types are   defined
as
monomorphic.

> message splitting, and type prediction) are used by the Self
compiler
> to
> achieve high performace.  In omega, elementary types are defined as

-------------------==== Posted via Deja News ====-----------------------
      http://www.dejanews.com/     Search, Read, Post to Usenet