Handle unknown b-value encoding

[kjg]

by treating it as ascii as well as ignoring any invalid chars

[kjg]

Quick question: How would you feel about replacing invalid chars with "�" instead of "". That way it is clear that something went "wrong" during translation where we otherwise just pretend everything was okay with no notice.

[jeremy]

Good call on � 👍

[kjg]

Should I change that in this PR or a new one?

Any idea offhand how to get Iconv to do � in 1.8.7?

If it isn’t easy / built in to Iconv, is it worth it to add for 1.8.7, or should I do it just for the String#encode versions?

[kjg]

Yeah, as far as I can tell, There is no good way to specify a replacement character in 1.8.7. Any 1.8.7 solution that I can think of to replace invalid characters would involve iterating over bytes and be really messy.

So the question becomes do we scrap � and leave the replacement as "" across the board, or do we allow the 1.8.7 output to be different.

I vote for allowing the 1.8.7 output to be slightly different.

[bf4]

@kjg Maybe https://github.com/bf4/encoded_string/blob/master/lib/encoded_string.rb#L14-L19 helps? I don't have 1.8.7 to test

[jeremy]

Separate PR makes sense.

[kjg]

This is passing now after a rebase. I'll make the replacement character change in a separate PR