During the initial capability exchange phase, the server picks a position encoding from the list offered by the client. To ensure communication is possible, both clients and server are expected to speak the "utf16 position encoding".
ocaml-lsp prefers UTF8 when available, but when it is not (which seems to be the case from VSCode), it should properly support utf16 positions. However, Merlin's utf8 positions are not migrated.
As a consequence, the error in
let x = {|a𐐀b|} ^ 3333333is reported in the wrong place (and semantic highlighting is wrong):
With something after to see the full wrong range:
During the initial capability exchange phase, the server picks a position encoding from the list offered by the client. To ensure communication is possible, both clients and server are expected to speak the "utf16 position encoding".
ocaml-lsp prefers UTF8 when available, but when it is not (which seems to be the case from VSCode), it should properly support utf16 positions. However, Merlin's utf8 positions are not migrated.
As a consequence, the error in
is reported in the wrong place (and semantic highlighting is wrong):
With something after to see the full wrong range: