Le 29/08/2016 09:46, Jan Åge Langeland a écrit :
On 29.08.2016 02:09, Tim Wescott wrote:
A kludge would be to make it into two integers.
Each case may not be so complex to work around, but in total it seems
like Scilab 6 is not yet well prepared for int64 and uint64.
For instance %x and %o need to be handled too, functions like
dec2bin() and dec2hex() should work - etc.
Generally a lot of caution is needed when using these 64 bit
integers. Example:
--> int64(2^62+256+1) //loss of prescision
ans =
4611686018427387904
Yes, but this loss is expected, since 2^62+256+1 is performed with
decimal encoding.
It is a case similar to the classical 1+ %eps/2
It is not a bug.
For dec2*(): int64 encoding is already supported, and a clear error
message is yielded for input integers > 2^52
--> dec2bin(int64(2^52))
ans =
10000000000000000000000000000000000000000000000000000
--> dec2bin(int64(2^53))
at line 12 of function dec2bin (
SCI\modules\elementary_functions\macros\dec2bin.sci line 43 )
dec2base: Wrong value for input argument #1: Must be between 0 and 2^52.
So, there is still a limitation, but it is clearly stated. At least,
integers from 2^33 to 2^52 are newly supported :)
Samuel
_______________________________________________
users mailing list
users@lists.scilab.org
http://lists.scilab.org/mailman/listinfo/users
