Peter Bourgon

Ah, yes, and to just make it explicit, you wrote > Especially when [implementations] accept [the] `iIlLoO` mapping, as [is suggested in the] original Crockford's base32 decoding spec, Lexicographically sortable...

I'm also blocked on this issue for a project. Can I help move things forward somehow? > However, one behavior I'd like to keep is not just ignoring unrecognized keywords,...

My intuition is that the bug is in the example, not the code. WDYT?

Okay, I see what you mean. I think the underlying problem here is an error in the original API design, and that design error's interactions with the code in the...

> https://medium.com/@dotronglong/interface-naming-convention-in-golang-f53d9f471593 This guidance is incorrect.

Idiomatic Go interfaces are named by their method set, and are never prefixed or suffixed by `I` or `Interface`. > By convention, one-method interfaces are named by the method name...

One consistent execution of diskv for a given directory is a core requirement.

Again, events are _by definition_ non-deterministic. Applications can not and must not rely on determinism in event delivery.

Oh! Sorry if I'm confusing things, here. When I say events are by definition non-deterministic, I don't mean that events have been defined to be non-deterministic, as if deterministic events...

I agree that it would be enormously valuable. The problem is that it isn't possible, without sacrificing essential properties of the system. _edit_: You're of course free to disregard my...