Floating point overflow

Question

Floating point overflow

RonLinu opened this issue 2 years ago · comments

Hello Stefan!
I guess you probably know about this but here it is: TinyBASIC locks up when handling a number bigger than 1e38.
print 1e39
x = 1e39
will lock up the interpreter.
I admit this does not occur often, mostly because of a programming error.

On the other hand, underflow is handled correctly as it returns 'zero' which is ok.

Tested on an Arduino Mega 2560

Stefan Lenz · Answer 1 · Tue Dec 06 2022 13:25:39 GMT+0800 (China Standard Time)

will look into this soon. i rewrote the number code anyway

…

Sent from my iPhone

On Dec 5, 2022, at 10:24 PM, RonLinu ***@***.***> wrote: Hello Stefan! I guess you probably know about this but here it is: TinyBASIC locks up when handling a number bigger than 1e38. print 1e39 x = 1e39 will lock up the interpreter. I admit this does not occur often, mostly because of a programming error. On the other hand, underflow is handled correctly as it returns 'zero' which is ok. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.

Stefan Lenz · Answer 2 · Wed Dec 07 2022 02:20:40 GMT+0800 (China Standard Time)

Checked in new code in the TinybasicArduino folder of the repo that fixes the problem. Origin is mishandling of inf. The code I checked in is portable to gcc variants. Will ceck portability.

Stefan Lenz · Answer 3 · Wed Dec 07 2022 03:39:24 GMT+0800 (China Standard Time)

Fixed now in the main code and tested against the main platforms. Handle inf and nan.