Converting a string to a double problem

[TheAutomator]

Hello!
my first post here, I'll keep the description of the problem simple:

vbscript:

msgbox Cdbl("3.14") ' OUTPUT = '3.14'

purebasic:

Debug ValD("3.14") ; OUTPUT = '3.1400000000000001' wut...?

I'm writing an interpreter in PureBasic and converting tokens from a text file / string to the accurate numbers is a must...
Does anyone know how to do this correctly?

Regards!

[mk-soft]

A float or a double is a structured number

A smooth 1.0 does not exist. Look at wiki how a float or double is stored in memory.

[TheAutomator]

okay seems logical but then explain the vb example?
what magic was going on there?

[BarryG]

explain the vb example?

VB probably cleans up the output automatically. Internally, VB would get 3.1400000000000001 as well.

In PureBasic, you can set the number of digits to show after the decimal point like this:

Code: Select all

Debug StrD(ValD("3.14"),5) ; 5 = Show 5 digits after the decimal point.

what magic was going on there?

VB holds the user's hand.

[infratec]

To be fair:

MsgBox needs a string.
So you need to convert the double back to a string:

Code: Select all

Debug StrD(ValD("3.14"), 2)

[TheAutomator]

To be fair:

MsgBox needs a string.
So you need to convert the double back to a string:

Code: Select all

Debug StrD(ValD("3.14"), 2)

That's what 'debug' is doing in this example

[TheAutomator]

explain the vb example?

VB probably cleans up the output automatically. Internally, VB would get 3.1400000000000001 as well.

In PureBasic, you can set the number of digits to show after the decimal point like this:

Code: Select all

Debug StrD(ValD("3.14"),5) ; 5 = Show 5 digits after the decimal point.

what magic was going on there?

VB holds the user's hand.

Damn, this is not going to be easy.. now i also need to find out how to dynamically find out how much digits to show if the value is different

[TheAutomator]

I think this might be a workaround :

Code: Select all

double.s = "3.14"
Debug StrD(ValD(double), Len(StringField(double, 2, ".")))

[VB6_to_PBx]

I think this might be a workaround :

Code: Select all

double.s = "3.14"
Debug StrD(ValD(double), Len(StringField(double, 2, ".")))

your workaround example seems to work well

Code: Select all

EnableExplicit
Define.s double = "3.1415926535897932384626433832795"
Debug StrD(ValD(double), Len(StringField(double, 2, ".")))

[C87]

If I want to remain in control of the result I only use integer calcs.

[NicTheQuick]

I'm writing an interpreter in PureBasic and converting tokens from a text file / string to the accurate numbers is a must

I don't know what you really mean with "accurate numbers". If it is what I understand under "accurate numbers" then floats and doubles are nothing you should use. Especially for financial calculus you should use the smallest unit and store them as an integer. For example instead of storing 1.23 € as a float number with the unit €, just store 123 as an integer with the unit ¢ or €/100. This way you can not get weird values in between.

[TheAutomator]

I think this might be a workaround :
your workaround example seems to work well

Code: Select all

EnableExplicit
Define.s double = "3.1415926535897932384626433832795"
Debug StrD(ValD(double), Len(StringField(double, 2, ".")))

uhm, that's just numerical overflow for a double right there, try:

Code: Select all

Define double.d = 3.1415926535897932384626433832795

[TheAutomator]

I'm writing an interpreter in PureBasic and converting tokens from a text file / string to the accurate numbers is a must

I don't know what you really mean with "accurate numbers". If it is what I understand under "accurate numbers" then floats and doubles are nothing you should use. Especially for financial calculus you should use the smallest unit and store them as an integer. For example instead of storing 1.23 € as a float number with the unit €, just store 123 as an integer with the unit ¢ or €/100. This way you can not get weird values in between.

well, if you would make an interpreter, it has to read text and convert it to tokens right?
for example:

contents of a script file:

Code: Select all

test = 3.14
msgbox test

the interpreter makes tokens from it:

Code: Select all

"test" = identifier
"=" = operator
"3.14" = double -> covert to actual number in memory!
"crlf" = newline
"msgbox" = function
"test" = identifier

wouldn't it be weird if you execute this script and it shows a message box saying 3.1400000000000001?
I'm looking for a way to correctly do this...

Your suggestion is to store 3.14 as 314 and only add a dot and convert it to a float if you need to?
how would that work?

[mk-soft]

Not yet looked up how double values are stored (wiki)

It is your task to display the double value at the output with the desired number of decimal places

Code: Select all

dblVal.d = 1.101
Debug "Unformated"
Debug dblVal
Debug "Formated"
Debug StrD(dblVal, 3)

Wiki: https://en.wikipedia.org/wiki/Double-pr ... int_format

[BarryG]

vbscript:
msgbox Cdbl("3.14") ' OUTPUT = '3.14'

Does this help?

Code: Select all

Debug StrD(ValD("3.14")) ; Outputs 3.14 like VBScript