Stop evaluating macros

[SuperFola]

In the current implementation of the language and macro processor, every argument passed to a macro function is evaluated as a macro if possible.

However this is a problem for some nodes that ressemble macros but aren't, eg (@ list index) the index could be a node index in the list instead of an element index.

There are multiple solutions to this:

• add quoting, quasiquoting and unquoting (hardest solution) like Lisp and Scheme do it
• add a way to stop argument evaluation on demand (easy)

I went with the second solution, see: #472 and would like to know what you think about it

[metaleap]

quasiquoting and unquoting (hardest solution)

Not hard, steal mine which follows this =)

To be able to quasiQuote/unquote is the coolest for code (and other lists) constructions, can't even imagine having a macro facility on the one hand but not those on the other?..

[SuperFola]

To be able to quasiQuote/unquote is the coolest for code (and other lists) constructions

It is! For now I went with a construction that stops evaluation (badly named $paste, can't call it quoting as it would quote ($paste a) like lisp does but rather generate a symbol a that would be evaluated by the compiler), which achieve nearly the same?

I'll have a look, thanks for the link!

[metaleap]

I'll have a look, thanks for the link!

Since that's based on MAL it has 2 list types (list, vec) so don't let the "vec" stuff confuse ya there.

This is a "special form" aka lazy-eval'd-builtin so it can call eval/apply right in there, making for a trivial impl that works trivially for a tree-walking interpreter but is probably no solution for writing a Lisp compiler — whether to native code or custom byte-code such as yours. Unless you anyway do eval/apply at compile time for macro-running, then it might be applicable. If it isn't, then maybe that's what you meant by "hardest solution" and sorry for my initial cockiness 😁