The following program let f x y = () ;; let print_int n = Printf.printf "%d\n" n; n ;; let print_int_list n = Printf.printf "%d\n" n; [n] ;; (print_int 0)::(print_int_list 42) ;; Some (print_int 0, print_int_list 42) ;; (print_int 0, print_int_list 42) ;; f (print_int 0) (print_int_li…

The right-to-left ordering when consing should be intuitive (otherwise every additional element after the head results in a concat), but I understand the confusion re: tuples and thus arguments to functions. This goes back to the original construction of the ZINC interpreter. The topic is a recurri…

[image] StrongerXi: Still, why? OCaml is not Haskell, why didn’t the language designer reinforce these evaluations to a “more natural” order (one that resembles C, Java, Python, etc.)? What’s the design rationale behind this? Two comments, in addition to what have already been said in earlier…

For the record, I believe that life would be simpler for everyone if OCaml specified a left-to-right evaluation order and enforced it. Yes, in theory it’s better to encourage people to write code that does not rely on execution order, and it can give some optimization opportunities, but in practice …

Thank you so much for your explanation. I definitely agree with you, especially on using sub-expressions that both compute a value and perform a side effect is often regarded as unnatural and bad style I wonder if this is the fault of the standard calling convention from C, where args that can’…

this ends up causing subtle bugs that waste people time. Oh man I can relate to that so much. Hours and hours of debugging plus a hard lesson. a type system that reliably catches evaluation-order-dependent program and asks them to be rewritten with explicit sequencing That sounds really cool. …

[image] gasche: For the record, I believe that life would be simpler for everyone if OCaml specified a left-to-right evaluation order and enforced it. Which order would you pick for labelled arguments, though? [image] gasche: but in practice we don’t have effective tools to catch this mi…

[image] Maelan: Which order would you pick for labelled arguments, though? I would expect left-to-right in the order of application at the call-site.

Note: when we say “left-to-right”, we mean “in code writing order” (from the beginning to the end). If OCaml had an input syntax in a right-to-left-writing language (for example an all-arabic syntax), then it would naturally be right-to-left.

Yes, exactly that. It is a matter of pushing the arguments to the stack in the order expected by the calling convention. For fixed-arity functions, you could probably say that it would be enough to reverse the convention, however with variable-arity functions (which are pervasive in OCaml because of…

[image] gasche: in a future world where we don’t keep the bytecode compiler as part of the toolchain, I think we would go for left-to-right evaluation. Is the bytecode compiler really such an issue? I mean, just put the input code into A-normal form and that is it, you get left-to-right evalu…

[image] silene: I mean, just put the input code into A-normal form and that is it, you get left-to-right evaluation for free. Probably at the same time get call/cc also for free if a modified ANF is used. I guess the evaluation order surprises people because there’s no widely available langu…

Order of evaluation in application/tuple/...?

Learning

gasche March 27, 2021, 4:53pm 10

I would expect left-to-right in the order of application at the call-site.

1 Like

Topic		Replies	Views
Evaluation order Learning	6	2000	May 30, 2023
Tuple and function execution order Learning tuple	9	2463	February 11, 2020
Evaluation order considered important for newbs to learn early Learning	4	970	November 3, 2017
About the order of definitions Learning	33	2485	April 27, 2023
Why are OCaml function arguments evaluated right-to-left? Learning language-design	2	6394	November 30, 2017

Order of evaluation in application/tuple/...?

Related topics