2010a and later parsing of int64() and uint64() ?

In releases before 2010a, literal numbers inside int64() and uint64() calls were understood by MATLAB as meaning to parse the numbers as class double and to call the int64() or uint64() method on the resulting double. For example,

>> uint64(18446744073709551600)
ans =
 18446744073709551615

Starting with release 2010a, the literal numbers are instead fully valued at parse time and given appropriate type.

I do not have R2010a or later available, so I would ask that people investigate the limits of that parsing in the new versions and report back.

For example:

• is uint64(18446744073709551600+0) the same as uint64(18446744073709551600) ?
• ditto uint64(18446744073709551600*1) ?
• ditto uint64(+18446744073709551600) ?
• ditto uint64(18446744073709551600.) ?
• ditto uint64(18446744073709551600.1) ?
• ditto uint64(18446744073709551600e0) ?
• how about int64(2^64-15) ?
• is uint64(18446744073709551600/2) equal to uint64(9223372036854775800) ?

And more generally, is there any expression X whose absolute value does not exceed 2^53, for which

B = int64(X);

is not the same as

A = X;
B = uint64(A);

?

Answer 1

[Friedrich]

Hi,

tested with R2011a 64bit on Win7:

>> int64(18446744073709551600)

ans =

    9223372036854775807

>> uint64(18446744073709551600+0)

ans =

   18446744073709551615

>> uint64(18446744073709551600*1)

ans =

   18446744073709551615

>> 
uint64(+18446744073709551600) 

ans =

   18446744073709551615

>> 
uint64(18446744073709551600.) 

ans =

   18446744073709551615

>> uint64(18446744073709551600.1)

ans =

   18446744073709551615

>> uint64(18446744073709551600e0)

ans =

   18446744073709551615

>> int64(2^64-15) 

ans =

    9223372036854775807

>> uint64(18446744073709551600/2)

ans =

    9223372036854775808

>> uint64(9223372036854775800)

ans =

    9223372036854775800

[Walter Roberson]

Thank you!

It appears that if any character is included (including unary plus) that the expression is evaluated as double and that result is converted.

I would appreciate a test with unary plus and with unary minus (negative sign) for int64() and a value in an appropriate range such as 9223372036854775800.

Also, now that I think of it, I would like to inquire about int64(-9223372036854775808)

[Walter Roberson]

Bump. For completeness, I'd appreciate the last few tests I mentioned so that I am certain that I have understood the parser.

[Friedrich]

Hi,

sorry but I don't get your second part with the unary plus and minus. Could you please explain to me what I have to do?

int64(-9223372036854775808)

ans =

-9223372036854775808

[Walter Roberson]

Thanks. Could I ask you to check int64(+9223372036854775800) ?

[Daniel Shub]

R2011a 64bit on Linux

>> int64(+9223372036854775800)

ans =

9223372036854775807