Aditya Sirish

Looking at my last review, I think this isn't too far off. @neilnaveen could you take a pass and we can land this? I think it'd go really well with...

That's a good point, we overload the one type with both paths and scheme-based keys. I wonder if we need --root-key-path and --root-key (for gpg / fulcio). This would also...

I don't think $(cat path) is very user friendly. Also, for unencrypted keys, we don't want the contents to be in the shell history (if it records with this resolved).

> For public keys I don't see the problem Agreed. However, I was thinking in terms of consistency for the signing flow as well.

Also, please rebase to get #407 in.

Thanks @neilnaveen!

Latest commit tries danwakefield/fnmatch in place of doublestar. doublestar seems nice with things like being able to validate a pattern when a user adds it but fnmatch is closer (possibly...

Thinking aloud, but I think what we really want is just fnmatch? No need of ** then, fnmatch doesn't special case directory separaters. It operates strictly on strings as well.

I think fnmatch would meet the original expectation trivially. No need for `**`, `*` would match `/` supporting subdirectories as well.

Do you mean where `foo` has three subdirectories `a`, `b`, `c` and you want a pattern to only apply to `a` and `b` and not `c`? fnmatch doesn't make this...