Does your DSL little language need operator precedence?

Posted by ingve 4 days ago

1910Original

Comments

Comment by vrighter 1 day ago

You can ignore precedence in the grammar, and then use a pratt parser or shunting yard or something to parse the precedence.

But yes, it does need it, usually. And it's not a huge thing to implement. I usually implement it in the grammar, with inline node folding inserted for left associative operators, which gets me a very nice clean AST.

Comment by bitwize 2 days ago

Not if it's s-expression-based! (laughs in smug lisp weenie)

Comment by aleph_minus_one 2 days ago

Or, if the programming language uses infix binary operators:

Not if the programming language has evaluation order from left to right, e.g.

2+3*4

is evaluated as

(2+3)*4.

For example J uses this kind of evaluation.

Comment by Someone 2 days ago

J is APL-inspired, and APL is right-associative, so that would surprise me. https://www.jsoftware.com/help/jforc/preliminaries.htm#_Toc1... agrees with that, saying

“All J verbs (functions and operators) have the same priority and associate right-to-left. For example, a b + c is equivalent to a * (b + c), not (a * b) + c.”*

Your point about not needing operator precedence still stands, though.

Comment by aleph_minus_one 2 days ago

> J is APL-inspired, and APL is right-associative, so that would surprise me.

You are indeed right (it has been quite a long time since I experimented with J):

> https://www.jsoftware.com/help/jforc/preliminaries.htm (scroll down to "Order of Evaluation")

Comment by lmz 2 days ago

Smalltalk also.

Comment by fjfaase 2 days ago

Just a single function per level is sufficient for implementing both right and left association. I do not see the problem.

Comment by recursivedoubts 2 days ago

hyperscript has some operator precedence, but within a given general precedence level you have to explicitly parenthesize if you use different operators:

https://github.com/bigskysoftware/_hyperscript/blob/06f9078a...

this eliminates most practical precendence questions

NB: one thing that may strike people as strange is that the parse methods are on the parse elements themselves, I like to localize everything about a parse element in one place

Comment by aappleby 2 days ago

Yes, but it doesn't need any funny parsing trick to handle them. Just parse the whole statement as a list of expressions joined by operators, and then you can convert the flat list into a precedence-respecting tree with a few lines of code and an operator-to-precedence table.

Comment by childintime 2 days ago

Yes, it's as easy as that. Or check out Jonathan Blow on precedence.

The infamous dragon book convinced people to use the wrong tools and have the wrong mindset. It was a work of incompetence. There were no dragons, but the book itself.