Jakob Nybo Nissen

> FWIW, it feels like you're conflating type safety and stability Right, sorry, I was being imprecise. I use this package for type safety. I agree that for file operations,...

I also added a check to check whether the package repo is a relative path. This is tricky to check for, because frustratingly, there is no way to check if...

Maybe it's uncommon to want to use relative paths, but I wouldn't think so. I would expect most users who wants a local registry stores it in a file system...

Yes, it's likely I did something wrong. Here is what I did: I have a centralized registry called `SSI` in a directory `/path/to/Registry`. I then enter Julia and type: ```...

It is completely fine that the repo should be bare. The issue is that it is not created as bare using `LocalRegistry`, it is not documented that it should be...

> `create_registry` is intended to create a checked out registry so it shouldn't be bare. The repo argument should point to a bare upstream though. Ah. Okay, so you need...

I can make a PR if it's agreed that current behaviour is a bug, but since current behaviour is explicitly tested for, I guess some people think it's correct behaviour?

If the example with limit=3 will split indefinitely, then it is still buggy because it violates your assumption that it will do limit-1 split

> The parameter is to limit the output to limit elements, not limit splits, so split needs to do limit-1 non-empty splits Indeed. And would you not agree that it...

Ah, so I guess we disagree on whether split should try to do as FEW splits as possible to obtain `limit` elements (what I take it you believe), or as...