BBC BASIC for Windows
Programming >> BBC BASIC language >> Determining the FLOAT mode in use.
http://bb4w.conforums.com/index.cgi?board=language&action=display&num=1427211578
Determining the FLOAT mode in use.
Post by sveinioslo on Mar 24th, 2015, 3:39pm

I need for a proc to know the FLOAT mode in use, and what version of BBC, 5 or 6.
Can't remember to ever have read how to, searching YAHOO gave much interesting info, but no answer.

So i wrote this program to determine the version and float mode.
There is perhaps a better and safer way ?
Svein

Code:
      *FLOAT 40
      PRINT "floatmode is ";FNfloat
      *FLOAT 64
      PRINT "floatmode is ";FNfloat
      ON ERROR : PRINT "BBCv_5" : END
      *FLOAT 80
      PRINT "floatmode is ";FNfloat
      PRINT "BBCv_6"
      END

      DEF FNfloat
      LOCAL a,b,A%,B%,I%,J%,eq%,bytes%
      a=1 : a*=1.0# : b=1 : b*=1.0#
      A%=^a : B%=^b
      bytes%=5
      REPEAT
        eq%=0
        FOR J%=1 TO 10
          a=RND(1)
          FOR I%=0 TO bytes%-1
            B%?I%=A%?I%
          NEXT I%
          IF a=b THEN eq%+=1
        NEXT J%
        IF eq%=10 THEN
          IF bytes%=5 THEN =40
          IF bytes%=8 THEN =64
          =80
        ENDIF
        IF bytes%=10 THEN =0
        IF bytes%=8 THEN bytes%=10
        IF bytes%=5 THEN bytes%=8
      UNTIL 0
Re: Determining the FLOAT mode in use.
Post by rtr2 on Mar 24th, 2015, 5:05pm

on Mar 24th, 2015, 3:39pm, sveinioslo wrote:
There is perhaps a better and safer way ?

Your code seems to report the *FLOAT mode incorrectly in v6.00a. Here I get:

Code:
floatmode is 80
floatmode is 80
floatmode is 80
BBCv_6

If you want to find the true *FLOAT mode you can do so directly by testing the relevant flag bits at ?419, which is documented on the Wiki:

Code:
      CASE ?419 AND 3 OF
        WHEN 0: PRINT "*FLOAT 40"
        WHEN 1: PRINT "*FLOAT 64"
        WHEN 2: PRINT "*FLOAT 80"
      ENDCASE

But if what you want to know is the precision of a (suffixless) 'variant' numeric variable the most direct method is to use CALL:

Code:
      DEF FNprecision
      LOCAL M%,P%
      DIM P% LOCAL 10
      [OPT 0
      .M%
      movzx eax,byte [ebp+6]
      mov edi,[ebp+2]
      mov [edi],eax
      ret
      ]
      CALL M%, P%, a
      = P% * 8

That also tells you the version, because if the returned value is 40 or 64 it must be BB4W v5 and if the returned value is 80 it must be BB4W v6.

Richard.
Re: Determining the FLOAT mode in use.
Post by sveinioslo on Mar 24th, 2015, 7:21pm

Great, thank you!

The precision of a (suffixless) 'variant' numeric variable was exactly what i wanted.
I do not know, if knowing the version will be required, but just i case......
Forgot about the 'Interpreter internal variables' page, just found some other needed info. smiley

Svein
Re: Determining the FLOAT mode in use.
Post by rtr2 on Mar 25th, 2015, 06:47am

on Mar 24th, 2015, 5:05pm, g4bau wrote:
But if what you want to know is the precision of a (suffixless) 'variant' numeric variable

Here's an even more straightforward method, which works by finding out how much memory the variable occupies:

Code:
      DEF FNprecision
      LOCAL s{}
      DIM s{a}
      = DIM(s{}) * 8

Richard.
Re: Determining the FLOAT mode in use.
Post by sveinioslo on Mar 25th, 2015, 3:47pm

Nice smiley

Svein