[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Fcat and visiting the arguments

From: David Bateman
Subject: Fcat and visiting the arguments
Date: Mon, 5 Jul 2004 23:46:39 +0200
User-agent: Mutt/1.4.1i

Ok, since I wrote a little code to create a CATOP function, I started
trying to do the rest with the goal of making [] available to user
types. However, with the idea of reusing what had already been done
and taking it in easy steps, I was looking at the Fcat function and
trying to see how to rewrite this in terms of a CATOP function.

The problem I have is it makes sense that CATOP performs the complete
concatenation in a single go and be done with it. However, in Fcat the
outer loop is over blocks of the args and the inner over the number of
args. This allows cat_ra to be simplified to function over a single
block of contiguous memory.

Now I seem to remember that this the change 

2004-03-01  Petter Risholm  <address@hidden>
        * Array.cc (cat_ra): Speed up implementation by avoiding N-d indexing.

effectively changed the implementation from what I'd find the easiest
and most logical to deal with, to the one described above. The stated
reason was speed. However, similar things might be done with the
original implementation by not using the increment_index function
within a block of contigous memory. So the question is, whether I can
change things back to something similar to what it was while removing
the use of increment_index as much as possible? Peter, have you tried
something like that? Alternatively, any other ideas of how to treat the
unification of Fcat and "[]"?


David Bateman                                address@hidden
Motorola CRM                                 +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

reply via email to

[Prev in Thread] Current Thread [Next in Thread]