#Article
8 messages in this thread
You mentioned in your Benchmark review that NEW and DISPOSE had been removed
from the formal definition of Modula-2 by Wirth. Is there a document beyond
"3rd, Corrected…" that decribes some of this later formalization. Also is
Leon going to implement his own fp functions (like TDI has in AMIGAX.mod)?
I'd hope he doesn't continue to go to disk for libraries.
What do you mean go to disk for libraries? Since he uses FFP he must go to
disk. That's the only disk-resident lib, so BM does not open it implicitly;
you gotta do it yourself. What's the big deal? I'm not aware of any other
document other than "3rd edition…" by N. Wirth that discusses the removal
of NEW and DISPOSE. There's nothing to it either; simply use Allocate and
Deallocate from the Storage module (or the Amiga-provided AllocMem/FreeMem &
AllocRemember).
– Steve –
Some rather learned Amiga friends pointed out that using ffp can be a real
plus in that it may be re-written to take advantage of a math co-processor if
ones out there. So, based on that, using the disk lib is certainly
beneficial. About the NEW/DISPOSE: Upon rereading some of "..3rd Edition.."
it discovered that Wirth had not included those "standard" functions as part
of the final defintion. So on both counts mentioned I stand corrected.
(humble posture)
Yup, that's the advantage of using the supplied math libs. Another advantage
is that the compiler writer (Leon) didn't have to implement the math routines
themselves (other than to code the interface to the library).
– Steve –
TDI used some routines for single precision floats that were originally
written in about 1982. The routines for the doubles were taken from some code
for the MAC. Just thought I'd mention those things. I don't know if there is
any real advantage to using IMPORTed math routines like that or not. I do
understand that IEEE maintains a greater level of accuracy. I'd think it best
to be able to option which (FFP or IEEE) one to use.
I think the 3rd edition of Wirth's book gets rid of NEW and DSIPOSE. As for
FP libraries, I'd rather have Leon use the disk libraries. This will make my
Mandelbrot programs work faster when I install the Floating Point
co-processor and replace the fp library. Otherwise, Leon would have to write
a whole FP support package for different kinds of co-processors….Richie
RE: Message to Steve F. One of thes days I will learn to thinkith before I
speakith. By the way – liked your review. In spite of the fact I have the
beta version of BM right here in front of me there are lots of things I don't
know about. Time to look into it a litle closer.
(* ET *)
I'm glad you like the review….You should try Benchmark. It's really a
superior product…if you start using it, after a while you;ll wonder how you
got along without it…..Richie 🙂