>1) Where does the bloat come from and under what conditions?
The bloat problem was fixed several CW revisions ago.
>4) With CW, does one have to use the FplAdd() type functions or do you just
use x*y, x/y, etc?
You don't need to include any .h files to use floating point or its
operators with the current compiler versions unless you wish to use the old
float manager functions.
>2) The only thing I can see that NewFloatMgr gives is the ASCII to Double
and Double to ASCII conversions. I'm sure I'm wrong on this.
If you need to convert a double to ASCII, this is the only way (unless you
code it yourself) Plus the conversion functions only do scientific notation.
>I'm trying to see if I would be better off going to Cw. I know there
>might be other reasons to use CW but I'm only interested in the handling
>of Floating Points (doubles) for this discussion.
I don't use GCC so I can only tell you adding floating point support to a CW
project does not add much overhead to your app.
-----Original Message-----
From: Michael S. Davis <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED] <[EMAIL PROTECTED]>
Date: Tuesday, September 07, 1999 9:13 AM
Subject: FP revisited for the umteenth time. Question
>
>I have read much of the archived data on using Floating Point with
>GCC (and CW). I have read that if one uses the native FP operations of
>the compiler that code gets bloated. I am using GCC right now and don't
>use the FplAdd, FplMult, or any other such function. Code does not seem
>any larger than other similar apps that I have seen.
>
>I'm confused over this whole thing, still.
>
>I did a simple test. I created a basic framework app that thad no math
>operations but did have x and y declared as doubles. I compiled and
>executed the program to see that it worked. I did not use NewFloatMgr.
>
>Then I perfromed an addition of x and y (x+y not FplAdd(x,y)) then
>recompiled and comparred the code sizes. The code size was only
>slightly different.
>
>1) Where does the bloat come from and under what conditions?
>
>2) The only thing I can see that NewFloatMgr gives is the ASCII to
>Double and Double to ASCII conversions. I'm sure I'm wrong on this.
>
>3) The Double To ASCII conversion is limited to e99 exponents. If one
>uses CW, is there a similar limit to conversions; that is limiting the
>exp to e99?
>
>4) With CW, does one have to use the FplAdd() type functions or do you
>just use x*y, x/y, etc?
>
>I'm trying to see if I would be better off going to Cw. I know there
>might be other reasons to use CW but I'm only interested in the handling
>of Floating Points (doubles) for this discussion.
>
>
>----------------------------------------------------
>Shoot-to-Win
>
>Protect the 2nd Amendment
>----------------------------------------------------
>
>
