package ppxlib

evar id produces a Pexp_ident _ expression, it parses its input so you can pass any dot-separated identifier, for instance: evar ~loc "Foo.bar".

Same as pexp_apply but without labels

pstr_value_list ~loc rf vbs = pstr_value ~loc rf vbs if vbs <> [], [] otherwise.

• deprecated [since 2016-10] use Nonrecursive on the P(str|sig)_type instead

unapplied_type_constr_conv is the standard way to map identifiers to conversion fonctions, for preprocessor that creates values that follow the structure of types. More precisely, path_conv path (sprintf "sexp_of_%s") is:

• sexp_of_t if path is "t"
• A.B.sexp_of_foo if path is "A.B.foo"
• A.B.sexp_of_f__foo (module A1) (module A2) if path is "A.B.F(A1)(A2).foo" type_constr_conv also applies it to a list of expression, which both prevents the compiler from allocating useless closures, and almost always what is needed, since type constructors are always applied.

Tries to simplify fun v1 v2 .. -> f v1 v2 .. into f. Only works when f is a path, not an arbitrary expression as that would change the meaning of the code. This can be used either for cleaning up the generated code, or to reduce allocation if f is a local variable (the compiler won't optimize the allocation of the closure).

Eta-reduction can change the types/behavior in some corner cases that are unlikely to show up in generated code:

• if f has optional arguments, eta-expanding f can drop them
• because labels commute, it can change the type of an expression: $ let f ~x y = x + y let f2 = fun x -> add x;; val f : x:int -> int -> int = <fun> val f2 : int -> x:int -> int = <fun> In fact, if f does side effects before receiving all its arguments, and if the eta-expansion is partially applied, eta-reducing could change behavior.

eta_reduce_if_possible_and_nonrec is meant for the case where the resulting expression is going to be bound in a potentially recursive let-binding, where we have to keep the eta-expansion when rec_flag is Recursive to avoid a compile error.

This module contains updated versions of node constructors that were kept stable when the node changed. For every function in this module, there's an equally-named function outside this module. The function outside this module will stay stable, whereas the function inside this module will adapt potential upcoming new compiler features. Only use a function in this module, if the equally-named one outside this module is missing a feature you need.