github doesn't always use hashes for gists

[s-u]

This rather a heads-up: github also uses non-hash references to gists. When you create a public gist on current GHE it has a short integer reference instead. This impacts discoverability and code that assumes gists IDs are hashes (e.g. where we distinguish a notebook name from the id).

[cscheid]

The same thing happens on public github, fwiw.

[cscheid]

(In what sense does this impact discoverability, though?)

[s-u]

It's much easier to guess the gist name so we cannot hide it anymore.

[cscheid]

Ah. Let's get our terminologies in sync here: I've been using "discoverability" as a positive thing, in the sense of "RCloud enables discoverability of work in your organization".

You're concerned about leaking secrets. You're right, but 1) that's by design on github, 2) that's why the default setting for create_gist is public=false: https://developer.github.com/v3/gists/#create-a-gist

How did this become a problem for you?

[gordonwoodhull]

Yeah, these are public gists, therefore not hidden.

[gordonwoodhull]

I think the thing to document here would be the three different kinds of public/private:

• GitHub public gist (this one)
• rcloud visibility/discoverability
• rcloud publish

[s-u]

@cscheid it has never been a problem for me, I don't care :)

@gordonwoodhull yes, I agree, this should be probably documented somewhere.

My main point here is that, technically, we cannot assume hashes.

[cscheid]

Does this trigger a bug somewhere in RCloud right now?

[s-u]

I'm pretty sure such notebooks cannot be used in notebook.R by ID, because the distinction between by-id and by-name calls is whether it has a hash form or not:

## is this first part a notebook hash?
if (grepl("^[0-9a-f]{20}$", pex[1L])) {

[cscheid]

@s-u, great. I'll open an issue for that.