MatChar equality is now based only on character content

[nedtwigg]

Possible resolution of #21

I think it was smart to separate the peculiarities of MATLAB's fileformat vs the actual semantic data in the format vs types package as you have done. format.CharEncoding really is Mat5 specific - I hadn't realized the issues around endian-ness and 16-bit ASCII. Because it's Mat5 specific, I don't think it makes sense to move it to types, which means we can't really add it to Char either.

What do you think about this little cheat?

[ennerf]

On second thought, do you have an actual use case for retrieving the backing encoding of an existing char array?

After doing a quick search I didn't find any equivalent functionality in MATLAB. native2unicode and unicode2native only convert to and from binary representations.

[nedtwigg]

If it's not possible to access the encoding in MATLAB, then I guess I don't :) And in that case, I should adjust the implementation of .equals() and .hashCode() to be encoding-agnostic, yeah?

[ennerf]

I think it should be ok to just check the equality for the char[] array. The encoding is only used for converting to and from binary data, so it has no impact on the data once it's loaded.

The only use case I can think of is when someone loads an existing .mat file and modifies a contained CharArray with unicode characters that are not covered by the previous encoding. In that case something like a setter or an automatic upgrade to a higher encoding would probably make more sense. It's such a corner case though that I'd be happy to ignore it until someone files an issue.

[ennerf]

Thanks