Create a Pcase pattern that works like -let.

[okamsn]

This PR is the beginning of a pcase pattern that works like -let. For me, this would resolve #376 (which I opened).

Before I copy the tests for -let and work through the edge cases of the pattern, would you be willing to merge a pcase pattern that works like -let?

[okamsn]

This is now ready for review.

I've copied the tests of -let for testing the pattern in pcase-let and pcase, except for a test of parallel bindings for pcase, which does not apply in that context.

Should the &hash? for only match hash table, or should it match if the value is nil?

Do you want any changes in how the code works or its style?

Thank you.

[okamsn]

I have updated the code to use dash--match-make-source-symbol when gensym is not available.

How would you like to handle Emacs versions when pcase-defmacro is not defined? It was added in Emacs 25.1. I put the code in an if expression with a warning if pcase-defmacro isn't bound. Would you like it done some other way?

[okamsn]

Another option would be making dash-pcase.el a separate package.

@basil-conto, do you have an opinion on how to handle pcase-defmacro not being defined in Emacs 24?

[basil-conto]

Another option would be making dash-pcase.el a separate package.

I don't really have an opinion on this; either way would work for me (I'm also quite unfamiliar with this part of Dash).

@Fuco1, @magnars any thoughts?

@basil-conto, do you have an opinion on how to handle pcase-defmacro not being defined in Emacs 24?

IIUC this feature is about hooking Dash destructuring into pcase, when possible.

IMO if it is not possible, Dash does not need to do anything: users running Emacs 24 should neither expect Dash to invent a pcase feature that pcase does not itself provide; nor should they see warnings about Dash not being able to hook into pcase.