OpenEuphoria

2. Re: ? 2e100_000 -- a killer statement

• Posted by mattlewis (admin) Jun 26, 2013
• 1446 views

? 2e100_000 --BAD 

? 2e100_00 -- OK 

It's the brute force way we parse scientific notation. It's done by std/scinot.e. 2 * 10^100,000 results in a lot of calls to carry() and a lot of sequence manipulation. It doesn't work well with such large exponents.

Matt

4. Re: ? 2e100_000 -- a killer statement

• Posted by ArthurCrump Jun 26, 2013
• 1347 views

? 2e100_00 -- OK 

Are you running a 64-bit Euphoria with a much bigger range of numbers? I am running 32-bit Euphoria 4.05 and ? 2e100_00 gives an answer in a reasonable time but I wouldn't say it's OK because it's the wrong answer. I get 1.412967193e+136 instead of 2e+10000; completely wrong.

Anything above about 1.8e308 gives a silly answer, sometimes with a negative exponent. For example, ?2e309 gives 6.180692095e-308

The numeric input routine should be able to detect that the number is out of range and report an error.

Arthur

6. Re: ? 2e100_000 -- a killer statement

• Posted by _tom (admin) Jun 26, 2013
• 1302 views

The reason for the silly number was that I was trying to crash Euphoria. I wanted to compare the error message that Eu gives compared to an overflow error message from Python. It all comes down to comparing error messages so I can make the claim that Euphoria error messages are more meaningful (friendlier) than Python.

For something basic, like an undefined variable, Euphoria messages are easier to understand.

TOM

8. Re: ? 2e100_000 -- a killer statement

• Posted by ArthurCrump Jul 04, 2013
• 1195 views

? 2e100_00 -- OK 

I have created a ticket for this Ticket:877

Arthur