Some suggestions/questions

[amterp]

First of all, really cool project! I've gotten really into writing useful bash scripts in the past year and as I've been getting into more advanced/complex things, I've found myself wishing for an alternative, and Amber is looking really promising!

I've been playing around with it and reading the onboarding docs a bit. I had the following questions/suggestions:

1. Q: Is there a reason to use exit from std over fail? It appears to me like you can use fail with unsafe if you wanted to replicate exit-like behavior. I guess the difference is that with exit, you keep the 'failure-scenario' internal to the function, rather than advertise the fail-ability of it. I can't think of a reason right now for when I'd want that, but I suppose it might exist?

2. Q: What is the purpose of needing main in order to allow fail / ? syntax? Could we not implicitly surround the entire run script in a main, removing the need to wrap your code and indent it (styling)? There's probably some language design/philisophy I'm not familiar with since it's not necessarily unique to Amber, but my initial impression is that it seems like an extra bit of unnecessary cruft to writing scripts.

3. Suggestion: when I got to the basic_syntax/commands (and the linked basic_syntax/functions) sections of the doc, I was confused about why my fail / ? attempts were failing to compile (with Fail statements can only be used inside of functions or the main block). By this point, the docs had not yet mentioned (to my memory) what a main block was, or that it was necessary (keep in mind I'm in a bash headspace when trying out Amber so it didn't click for me). Only until basic_syntax/importing does the main block get mentioned/explained. I'd recommend mentioning that earlier in the docs, since it seems like basic pre-requisite knowledge to understanding fail / ? :)

4. Suggestion/discussion: I've seen some discussion of this online as well on Amber, and I think I lean towards splitting Num into Int and Float making sense. In most cases, my scripts are only gonna be dealing with Ints, so the compiled bash paying the cost of needing to handle floating points feels a bit wasteful. Idk if you guys have plans/ideas for how to optimize this away while keeping the Num type, but interested to hear where people's thinking is on this :) Just saw issue 70, ignore!

Thanks again for your work on Amber, really exciting project, best of luck with it!

[amterp]

Another question I forgot to ask:

Is ease of arg parsing seen as something that would be fair game for Amber to tackle? Even if via std utils? Or is it seen as more of a '3rd party library' responsibility / push onto Amber script writers? I'm referring to things like parsing arguments/flags.

[b1ek]

yeah, i think that we will have a powerful argument parser in stdlib sometime. probably not very featureful, but useful nonetheless

[b1ek]

attempts were failing to compile (with Fail statements can only be used inside of functions or the main block)

that's a design/docs oversight, open an issue to discuss it there

[Ph0enixKM]

Thank you so much @AlexanderTerp! Here are the answers:

1. A: I was thinking about removing exit from the library but then I realised that people would be using fail 0 for the reason of exiting with successful exit code - that syntax appears to me a bit misleading so I eventually left it.
2. A: It's for the reason of writing (for example) testing code in the main file but not running it when writing a library.
3. A: Thank you for this suggestion. Could you please create an issue here for that? https://github.com/Ph0enixKM/Amber-Docs