Chris de Graaf

I'm not positive on this, but I don't think that this would hurt reproducibility at all. I think that without a manifest checked in, the dependencies are resolved from scratch...

Have you tried using `copyto!` as recommended?

It should be simpler, just doing a `get_repo` on the upstream ought to give us the default branch. I'll look into this tomorrow. edit: s/tomorrow/EVENTUALLY/

My guess is that GitHub.jl or HTTP.jl is failing to URL-escape the path.

I think most of its uses can be replaced, but not trivially. There is one instance of mocking `now` where that test would just need to be removed.

Ah interesting. I think the rate limits for Actions are the same as for Apps (5000/hr). For old versions especially, trying to look up the registry PR uses a lot...

[They're using a Documenter key as the GH token](https://github.com/qojulia/CollectiveSpins.jl/blob/52f34e3c995edd4797b2b7a9958c59c571d8fab0/.github/workflows/tagbot.yml#L14) User error, but there's indeed a reporting bug: we can't check if the repo is private with bad credentials. I'm leaning...

I'll paste what I put in [Discourse](https://discourse.julialang.org/t/ann-the-tagbot-github-app-is-deprecated-in-favour-of-the-tagbot-github-action/34344/27?u=christopher-dg) just now: I would have liked to have written it in Julia,. But I’ll be totally honest: Python is objectively better for this...

I think I'm ready to give this a shot. With #173 and #157 having basically no way forward that doesn't involve huge refactoring or technical debt (I chose the latter...

Bump, I could use this. FWIW, I haven't had any Taiko issues with osu-bot.