dojox.math.BigInteger not usable with base 10

[rdunklau]

I tried to use dojox.math.BigInteger to perform client-side checksum calculation, and I stumbled upon several bugs.

Steps to reproduce: Executes the attached code. Note that it succeeds at instantiating bigints in base 16, but not in base 10.

Solutions: It seems that some functions are not properly exported from BigInteger to BigInteger-ext. For example, the fromRadix method is accessed as if it were public in BigInteger, although it is only defined in BigInteger-ext and as a protected method (named _fromRadix). Same goes for the intAt function, defined in BigInteger, and not exported in the namespace to be used by BigInteger-ext.

I'm willing to submit a patch and a CLA for these.

[bill]

Looks like Eugene owns this file?

[Eugene Lazutkin]

I'm willing to submit a patch and a CLA for these.

Please do both.

[hreusser]

Hi rdunklau,

is it possible to provide the patch? We are struggling with the same issue. Many thanks in advance.

For general curiosity: Why is the big int library splitted up into two modules?

-thx Halm

[cyan]

The following was failing for me (on dojo-release-1.8.0b1):

require([
    'dojox/math/BigInteger',
    'dojox/math/BigInteger-ext'
], function(BigInteger) {
    var i = new BigInteger("1234", 10);
});

with the message "Object 0 has no method 'fromRadix'".

I fixed the issue by fixing the call to fromRadix in function bnpFromString of BigInteger? (it was missing a leading _):

else { this._fromRadix(s,b); return; }

It looks like the same issue as on this ticket.

I signed a CLA years ago if it helps (don't think a 1 character bugfix can really have copyright issues anyway).

[bill]

In [29447]:

fix apparent typo, this should have tests though, fixes #12174 !strict, thanks cyan