Consider syntax for running hegel outside of test code

[Liam-DeVoe]

#[hegel::test] currently automatically adds the #[test] annotation, which means it is conditionally compiled only under cfg(test). But there are valid reasons to run hegel outside of test code, and we should support this.

Options:

• (1) Add a new #[hegel::main], which is identical to #[hegel::test] except it does not add #[test].
• (2) Change #[hegel::test] to not add #[hegel::test] automatically.
• (3) Add a parameter to #[hegel::test] which disables the automatic #[hegel::test] annotation.

[DRMacIver]

I have a strong preference for #[hegel::main]. I think we're going to want it eventually for antithesis workloads, and having a first class main entry point lets us expose a good CLI for configuring hegel behaviour (e.g. we can automatically add command line args for number of tests, verbosity, etc)

[Liam-DeVoe]

To clarify: the CLI you're imagining is something we build on top of hegel, right, not something provided by #[hegel::main] by default? I don't mean to imply with #[hegel::main] that the decorated function be the single entrypoint to the program. You can for example easily have multiple #[hegel::main] functions, and choose how to call them. It really is identical to #[hegel::test] without the #[test] attribute.

[DRMacIver]

To clarify: the CLI you're imagining is something we build on top of hegel, right, not something provided by #[hegel::main] by default? I don't mean to imply with #[hegel::main] that the decorated function be the single entrypoint to the program. You can for example easily have multiple #[hegel::main] functions, and choose how to call them. It really is identical to #[hegel::test] without the #[test] attribute.

I'm fine with having such a function but think we should not call it [hegel::main] if we do. main very much reads as "CLI entry point" to me.

[Liam-DeVoe]

Yeah agree.