suggestion for behaviour of mimetype_isa

[doriantaylor]

Hello,

I found myself writing the same mimetype_really_isa function over and over again, such that I eventually made it into a module of its own (a Role, actually):

https://github.com/doriantaylor/p5-role-mimeinfo/blob/master/lib/Role/MimeInfo.pm#L80

The code changes the behaviour of mimetype_isa the following ways:

• mimetype_isa('a/b', 'a/b') now returns true
• mimetype_isa('a/b.noncanon', 'a/b') now returns true
• mimetype_isa('a/b', 'x/y.noncanon') now returns true (assuming a/b is a x/y to begin with)
• mimetype_isa('a/b') will return the entire set of ancestor types (including a/b and all paths up to application/octet-stream)

The purpose was to create a mimetype_isa that would take as wide a range of input as possible and didn't need any finessing first, which I found myself repeatedly doing with my own ad-hoc mimetype_really_isa function.

So, I wrapped it up into a Role. It really shouldn't be a Role. Indeed the code should probably be in this module.

But, of course, the behaviour is subtly different, so I'm not about to do a fork/clone/PR without talking to you first.

Thoughts?