Peter Law

Huh, it looks like that changed since this was reported. The originally linked code was https://github.com/actions/toolkit/blob/b4639928698a6bfe1c4bdae4b2bfdad1cb75016d/packages/glob/src/internal-hash-files.ts#L16-L19 It seems that the equivalent check is still present https://github.com/actions/toolkit/blob/45c49b09df04cff84c5f336f07d5232fa7103761/packages/glob/src/internal-hash-files.ts#L23-L26 Though as @alexkuc notes...

@RealOrangeOne (semi randomly chosen) just wondering if you're able to help out here? Is there anything outside contributors can help with to progress this?

A change such as this, or possibly allowing a number of different styles, would avoid conflicts with Jinja 2, which treats `{%` as a special block. Jinja 2 is pulled...

That's ok. Do you have any suggestions on why this might be working in the ordinary diamond hierarchy? (i.e: the case with only two, not three, legs)

Hrm, removing `--autoresolve` also yields the result I'm after.

Figured out why removing `left.in` or `right.in` from `dev.in` works -- because then the autoresolve logic can't find a common "sink" file which references all the other environments. I think...

Changing the torch line to `torch == 2.0.0, != 2.0.0+cpu` does seem to help here, however this forces me to manually pin the version in the input. It also seems...

Not completely. Mostly as I don't understand what kind of thing the `+cpu` adjustment to the version number is. I'm _guessing_ that `pip-compile` doesn't have this issue since it uses...

@judahrand are you aware of mypy's [`--implicit-optional`](https://mypy.readthedocs.io/en/stable/command_line.html#none-and-optional-handling) behaviour? The `foo: str = None` style is valid when using that option. (I don't know if other checkers have a similar behaviour,...

@danpalmer as there is (now?) a prometheus plugin for routemaster, is this completed?