Status: New
Owner: ----
New issue 3512 by [email protected]: Inconsistent printing of denormal
floats
http://code.google.com/p/v8/issues/detail?id=3512
First, consider the following utility functions:
function _DoubleHi(f) {
// Return the most significant 32 bits of a double float number.
// This contains the sign, exponent, and 21 bits of the mantissa.
var buf = new ArrayBuffer(8);
(new Float64Array(buf))[0] = f;
// Index 1 if the machine is little-endian. Use index 0 for big-endian.
var hi = (new Uint32Array(buf))[1];
// Return as a signed integer
return hi | 0;
}
function _DoubleLo(f) {
// Return the least significant 32 bits of a double float number.
// This contains the lower 32 bits of the mantissa.
var buf = new ArrayBuffer(8);
(new Float64Array(buf))[0] = f;
// Index 1 if the machine is little-endian. Use index 1 for big-endian.
var lo = (new Uint32Array(buf))[0];
return lo;
}
function _ConstructDouble(high, low)
{
var buf = new ArrayBuffer(8);
// This following is for a little-endian machine. For a
// big-endian machine reverse the indices.
(new Uint32Array(buf))[1] = high;
(new Uint32Array(buf))[0] = low;
return new Float64Array(buf)[0];
}
function _DoubleBits(x)
{
return [_DoubleHi(x), _DoubleLo(x)];
}
Then run _ConstructDouble(262271,558054906). JS prints this as
5.565382338080105e-309
However, _ConstructDouble(262271,558054906) is the exact rational number
4505782028811770/2^1076. If I use maxima to print this out using 32 digits
of precision, I get:
bfloat(4505782028811770/2^1076); ->
5.5653802702119232229811010819437b-309
I would expect JS to print something closer to this value than what is
actually printed. When JS reads the values produced by maxima, we get
_DoubleBits(5.5653802702119232229811010819437e-309) ->
[262271, 139513726]
It appears, however that JS has print/read consistency since
_DoubleBits(5.565382338080105e-309) ->
[262271, 558054906]
It's just that the printed value isn't quite what I would expect.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
