package merlin-lib
Install
Dune Dependency
Authors
Maintainers
Sources
sha256=ec23f324f875520cd8897f303cc6d4e595f3d7000914d410729f16b86ad1d70e
sha512=8db22100cc0af65b08f456a2a7af84e75396f5869ee7552f1f5888a1c0279d1d85e6eecb3a677ae6f0973a99823cddba0563843ce216197255667342ef161885
doc/merlin-lib.ocaml_parsing/Ocaml_parsing/Extend_helper/index.html
Module Ocaml_parsing.Extend_helper
Source
Generate an extension node that will be reported as a syntax error by Merlin.
Physical locations might be too precise for some features.
For instance in: let x = f in y ^1 ^2
Merlin cannot distinguish position ^1 from ^2 in the normal AST, because IN doesn't appear in abstract syntax. This is a problem when completing, because a different environment should be selected for both positions.
One can add relaxed_location attributes to make some locations closer to the concrete syntax.
Here is the same line annotated with physical and relaxed locations: let x = f in y
-- physical locations for f and y nodes
-- relaxed locations for f and y nodes
If some code should be ignored by merlin when reporting information to the user, put a hide_node attribute.
This is useful for generated/desugared code which doesn't correspond to anything in concrete syntax (example use-case: encoding of some js_of_ocaml constructs).
The converse: when merlin should focus on a specific node of the AST. The main use case is also for js_of_ocaml.
Assuming <code> is translated to:
let module M = struct let prolog = ... (* boilerplate *)
let code = <mapping-of-code>
let epilog = ... (* boilerplate *) end in M.boilerplate
To make merlin focus on M.code
and ignore the boilerplate (M.prolog
and M.epilog
), add a focus_node
attribute to the M.code
item.
val classify_attribute :
Parsetree.attribute ->
[ `Other | `Relaxed_location | `Hide | `Focus ]