2

I'm trying to convert a 64 bit binary string value () in to an integer but I'm apparently hitting the tcl limit and getting:

integer value too large to represent

Binary String:

000000000000101000000000000010100000000000000100

The proc I'm using to convert from binary to integer is:

  proc binary2decimal {bin} {
     binary scan [binary format B64 [format %064d $bin]] I dec
     return $dec
  }

I tried using 64 for formatting and also tried LI to no avail. I read about "wide" but did not understand if it could be applied here.

P.S: I'm using tcl 8.4

egorulz
  • 1,455
  • 2
  • 17
  • 28

1 Answers1

6

First off, the error is coming from format, which isn't very happy with 64-bit numbers by default (for backward compatibility reasons) but we're really working with a string of characters there, so %064s (s, not d) is a suitable fix.

You really ought to switch to using Tcl 8.5 where this is much easier:

binary scan [binary format B64 [format "%064s" $bin]] W dec

Or even just:

set dec [expr "0b$bin"]

In 8.4, you've got to do more work as the 64-bit support was a lot more primitive.

binary scan [binary format B64 [format %064s $bin]] II d1 d2
set dec [expr {wide($d1)<<32 | wide($d2)}]

That's slightly non-obvious, but the I format is and always has been documented to work with 32-bit values only. So we stitch the two (big-endian) half-values back together after scanning.

Donal Fellows
  • 133,037
  • 18
  • 149
  • 215