Page 2 of 2
Posted: Sat Apr 05, 2003 11:27 am
by BackupUser
Restored from previous forum. Originally posted by geoff.
Geoff:
I don't believe we need floats and doubles
Tinman:
I think PB should keep floats even if it gets doubles.
Geoff:
Yes, we need both.
Even if expressions are evaluated using doubles, we need float arrays for
applications that use huge arrays but can manage with modest accuracy.
Posted: Sat Apr 05, 2003 12:02 pm
by BackupUser
Restored from previous forum. Originally posted by Manolo.
Hi geoff,
I make one change in you function for work with 2 decimals all time:
Code: Select all
Procedure.s StrG(a.f)
; Convert float to string with n digits precision
n=Len(Mid(StrF(a),1,FindString(StrF(a),".",1)+1))
If Int(a)=0
n-1
EndIf
If Int(a)-a=0
n=Len(StrF(Int(a)))+2
EndIf
a$=""
If ab:ProcedureReturn("O/F"):EndIf
b=1/b
If a7:n=7:EndIf;max float accuracy
b=1:For i.l=1 To n:b=b*10:Next i
e.l=0
While a>=b:a/10:e+1:Wend
b=b/10
While a=0 And e-n And e1:a$=a$+".":EndIf
a$+Right(Str(i),n-1)
i=e+n-1
If i<0
a$+"E-"+RSet(Str(-i),2,"0")
Else
a$+"E+"+RSet(Str(i),2,"0")
EndIf
EndIf
ProcedureReturn a$
EndProcedure
And work fine. Thanks for this extraordinary function.
Manolo
Posted: Sat Apr 05, 2003 4:43 pm
by BackupUser
Restored from previous forum. Originally posted by geoff.
Manolo,
The latest version of my procedure gives the option of n digits precision or n decimal places. This has some advantages over your modification to the old routine.
1) It allows variable decimal places.
2) It works over a wider number range
3) It does not use StrF(), which will likely change in the future.
Code: Select all
Procedure.s zstrf(a.f,n.l); For n>0 convert float to string with n digits precision
; For nb:ProcedureReturn("O/F"):EndIf
b=1/b
If aValF("2e9")
a$="Too long"
Else
i=Round(a+0.5,0)
b$=Str(i)
If n=0
a$+b$
Else
If Len(b$)8:n=8:EndIf;max float accuracy
b=1:For i.l=1 To n:b*10:Next i
e.l=0
While a>=b:a/10:e+1:Wend
b=b/10
While a=0 And e-n And e1:a$=a$+".":EndIf
a$+Right(Str(i),n-1)
i=e+n-1
If i<0
a$+"E-"+RSet(Str(-i),2,"0")
Else
a$+"E+"+RSet(Str(i),2,"0")
EndIf
EndIf
EndIf
ProcedureReturn a$
EndProcedure
Posted: Wed May 07, 2003 11:12 am
by Martin Schönegg
I do not aggree with the opinion that we need Str instaed of StrF... I know very well what kind of variable i use and there is no problem to use the correct command. But I need the scientific way to put values in and out of my prog.
My first Z80 based computer knows the command using and I need at least a similar way to display values.
USING "###.##" creates the sign, two digits, the decimal point and two digits behind
USING "##.##^" creates sign, one digit bevore and two after decimal point with scientific Exponent (always 2 digits plus sign= 4 characters)
64 bit floats would be great and may be we can live without the old 4-byte-floats. The only thing I can imagine to be useful with this old format is an Array with several dimensions of floats creating huge memoryneeds. But : The inflation with memory should give us legitimation to forget 4-byte-floats at least the focus should be changed.
Please give us double-precision floats and a practicable way to use them. Thanks
Martin Schönegg
I'm not a game programmer, but engineer that needs a tool to create fast and stable solutions. Pb semms to be on the right way to be this tool