[Coco] questions about constants

[jdaggett at gate.net]

Wayne 

They are basically the same. Prior to IEEE one method for a 32 bit FP number was to 
have the sign stored in its own byte. This left the mantissa at 24 bits, the exponent in the 
its byte. Therefore five bytes were needed to store a single FP number. The added 
resolution really is insignificant. What the IEEE single precision format does is to reduce 
the storage requirements by one byte. Back in the old 8 bit days where memory was 
precisous, that was a concern. 

james

On 13 Sep 2009 at 13:14, Wayne Campbell wrote:

> James,
> 
> I knew there was something different about Basic09 REAL variables and OS-9 C floats/doubles.
> 
> In Basic09, a real value is defined:
> 
> Byte 1: Exponent
> Bytes 2-5: Mantissa, with Byte 5, Bit 0 being the sign bit
> 
> which is almost the exact opposite of the OS-9 C float:
> 
> Bit 7 of Byte 1 being sign bit, and Bytes 1-3: Mantissa
> Byte 4: Exponent
> 
> I have never understood why that difference exists, but being prior to the IEEE standard would be as good an explanation as any.
> 
> I believe I understand the compiler directives and pseudo op codes better now. In order to make sure I am looking at things correctly, the following is true?
> 
> org 0 sets the current . value to 0
> 
> A.LABEL1 rmb 1 this A memory address is relative 0 to org
> A.LABEL2 rmb 1 this A memory address is relative 1 to org
> 
> A.LABEL3 rmb 2 this A memory address is relative 2 or org
> A.LABEL4 rmb 1 this A memory address is relative 4 to org
> 
> I know that equ is a system equate. What I'm not sure of, is what it is equating.
> 
> For example, say the above is true, and the following line comes next in the list:
> 
> A.LABEL5 equ 
> 
> does this mean that A.LABEL5 is being set to a value of 5, or the address of 5 relative to org?
> 
> Wayne
> 
> 
> 
> ________________________________
> From: "jdaggett at gate.net" <jdaggett at gate.net>
> To: Wayne Campbell <asa.rand at yahoo.com>; CoCoList for Color Computer Enthusiasts <coco at maltedmedia.com>
> Sent: Saturday, September 12, 2009 8:16:24 PM
> Subject: Re: [Coco] questions about constants
> 
> 
> 
> Wayne
> 
> FCB, FCC, FCS are assembler directives and not opcodes. What they do is tell the 
> assembler at that place in memory to reserve a single byte, a double byte or multiple 
> bytes. This used to hold a spot for data, screen prompts or table entries. 
> 
> Memory for all processors are in binary form. Integers are represented as a binary 
> number. They can be 8 bit, 16 bit or even 32 bit form. The more bits used for integer the 
> larger teh integer value you can have. 8 bit integers are limited 0 tp 255 if unsigned and -
> 127 to 128 if signed. 
> 
> Real numbers are stored as floating point binary numbers per the IEEE standard. Well 
> the Coco is not IEEE standard as the DEC Basic was written before the IEEE standard 
> was formed. To store a IEEE binary floating point number you need to 4 bytes for single 
> precision, 8 bytes for double presicion. In single presicion you have a 23 bit mantissa, 
> one sign bit and an 8 bit weighted exponent. For double presicion you have an 11 bit 
> weighted exponent, one sign bit and 62 bit mantissa. 
> 
> These require any program to reserve appropriate amount of memory for each number 
> used. You can use FCBs to reserve the bytes needed. In the case you only want to 
> declare the space and not assign a value,  you can use the asembler directive RMB to 
> reserve space. If you wish to reserve 4 bytes you can do this.
> 
> FPACCA    RMB     4
> 
> This will reseve 4 bytes for floating point accumaltor A. 
> 
> Assembler directives are a way to tell the assembler that all the code is not machine 
> code. Some maybe data, a macro so that you do not have to type the same code over 
> and over again. 
> 
> hope this helps
> 
> james
> 
> On 12 Sep 2009 at 16:41, Wayne Campbell wrote:
> 
> > I have looked at a dozen tutorials on assembly language programming. None of them address the op codes fcb, fcc or fcs, so I'm asking the assembly gurus on this list to help explain this to me.
> > 
> > What are the differences between fcb, fcc and fcs?
> > Is there a fci or fcr(/f/d) (for integer and real(/float/double) values)?
> > 
> > I know that:
> > 
> >   fcb = form constant byte   = any constant numeric byte value?
> >   fcc = form constant char   = any constant character that is displayable?
> >   fcs = form constant string = any constant string?
> > 
> > Are the following equivalent, so far as the assembler is concerned?
> > 
> > A. form constant byte:
> > 
> >   fcb   $20  
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $20 
> >   fcb   $42 B
> >   fcb   $41 A
> >   fcb   $53 S
> >   fcb   $49 I
> >   fcb   $43 C
> >   fcb   $30 0
> >   fcb   $39 9
> >   fcb   $0A  
> > 
> > B. form constant string:
> > 
> >   fcs   /            BASIC09/
> >   fcb   $0A
> > 
> > C. would using form constant char be equivalent?
> > 
> >   fcc   ' '  
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   ' ' 
> >   fcc   'B'
> >   fcc   'A'
> >   fcc   'S'
> >   fcc   'I'
> >   fcc   'C'
> >   fcc   '0'
> >   fcc   '9'
> >   fcb   $0A
> > 
> > Wayne
> > 
> > 
> > 
> >      
> > 
> > --
> > Coco mailing list
> > Coco at maltedmedia.com
> > http://five.pairlist.net/mailman/listinfo/coco
> 
> 
>       
> 
> --
> Coco mailing list
> Coco at maltedmedia.com
> http://five.pairlist.net/mailman/listinfo/coco