decoding no longer fails when unrepresentable tag size is encountered - #10

Hi, Thanks for you PR.
Could you please update the unit tests so travis is happy?

I should really start using these unit tests.

Just commented out the greater than max representable for now. All Matroska says is that an unknown element size should be represented by all 1's, so I can't imagine there would be a need to use more than 8 bytes.

Were you wanting a new unit test for a sample livestream? I could probably make a Big Buck Bunny livestream sample.

You could add a test that checks if the return value for readVint equals { length: x, value: -1 } so we would see if this expected behavior changes in the future.
Should be as simple as this:

it('put a clever message here', function() {
    readVint(new Buffer([0x01, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00]), -1);
});

Have a look at the (bad named) readVint test function here: https://github.com/siphontv/node-ebml/blob/master/test/ebml.js#L7

Good call. Let me know if you want any other changes (or a more clever message).

👍

Sweet! Thanks for the merge.

I'm pretty new to node. Do you have to update the version for npm? npm install doesn't include the changes. Maybe 0.2.1?

Oh, right. Well. Since this is a change in the api 1.0.0 would be correct, wouldn't it?

Okay, published to npm as [email protected]

Should lines 10 and 11 of lib/ebml/tools.js have been modified/removed as well?