Re: GNAT: Performance of String functions

[dewar@merv.cs.nyu.edu (Robert Dewar)]

Jakob says

<<Well, your doubts fails you. It is the SAME algoritm, it has been
translated
mechanically, and modified in minimum way to use ada functions were
mechanically, and modified in minimum way to use ada functions were
applicable.

But the overall algoritm of the translated program is the same. The
VAX-pascal =

is also modular so things are wrapped in packages the same way as in the
old code.
Certain Pascal concepts have been exchanged with ADA equivs, systems
calls are
of course new. One of these is strings and strings comparison. One
mistake, perhaps =

is that vax-strings have been replaced with unbounded strings instedad
of bounded s'.
>>


You missed my point, yes of course you can translate the high le vel
structure of the algorithm, but what interests us here is the algorithms
and data structures used for string handling, and what I meant when I
said that I doubted you could do the same thing in Pascal, was that
I doubted you could duplicate the algorithms and data structures of
unbounded strings.

And indeed that is the case, the proper translation is to bounded strings.