Improve configuration file parsing and CC backend compiler detection

lukstafi · lukstafi · commit 6e022aaba90f · 2025-02-24T11:33:08.000+01:00
- Add support for empty config values, which now use default settings
- Modify config file parsing to handle duplicate keys and strip prefixes
- Update CC backend to dynamically detect compiler from OCaml configuration
- Update example config file with clarified comments about configuration behavior
diff --git a/arrayjit/lib/cc_backend.ml b/arrayjit/lib/cc_backend.ml
@@ -15,7 +15,24 @@ let name = "cc"
 let optimization_level () =
   Int.of_string @@ Utils.get_global_arg ~default:"3" ~arg_name:"cc_backend_optimization_level"
 
-let compiler_command () = Utils.get_global_arg ~default:"cc" ~arg_name:"cc_backend_compiler_command"
+let compiler_command =
+  let default =
+    lazy
+      (let ic = Unix.open_process_in "ocamlc -config" in
+       let rec find_compiler () =
+         match In_channel.input_line ic with
+         | None -> "cc" (* Default fallback *)
+         | Some line ->
+             if String.is_prefix line ~prefix:"c_compiler: " then
+               String.drop_prefix line 12 (* Length of "c_compiler: " *)
+             else find_compiler ()
+       in
+       let compiler = find_compiler () in
+       ignore (Unix.close_process_in ic);
+       compiler)
+  in
+  fun () ->
+    Utils.get_global_arg ~default:(Lazy.force default) ~arg_name:"cc_backend_compiler_command"
 
 module Tn = Tnode
 
diff --git a/arrayjit/lib/utils.ml b/arrayjit/lib/utils.ml
@@ -115,7 +115,7 @@ let filename_of_parts = function
 
 let config_file_args =
   match read_cmdline_or_env_var "no_config_file" with
-  | None | Some "false" ->
+  | None | Some "false" -> (
       let read = Stdio.In_channel.read_lines in
       let fname, config_lines =
         let rev_dirs = List.rev @@ filename_parts @@ Stdlib.Sys.getcwd () in
@@ -141,13 +141,19 @@ let config_file_args =
                    lowercase @@ strip ~drop:(fun c -> equal_char '-' c || equal_char ' ' c) key)
                in
                let key =
-                 if String.is_prefix key ~prefix:"ocannl" then String.drop_prefix key 6 else key
+                 if String.is_prefix key ~prefix:"ocannl" then
+                   String.drop_prefix key 6 |> String.strip ~drop:(equal_char '_')
+                 else key
                in
-               Some (String.strip ~drop:(equal_char '_') key, v)
+               if String.is_empty v then None else Some (key, v)
            | l ->
                failwith @@ "OCANNL: invalid syntax in the config file " ^ fname
                ^ ", should have a single '=' on each non-empty line, found: " ^ String.concat l)
-      |> Hashtbl.of_alist_exn (module String)
+      |> Hashtbl.of_alist (module String)
+      |> function
+      | `Ok h -> h
+      | `Duplicate_key key ->
+          failwith @@ "OCANNL: duplicate key in config file " ^ fname ^ ": " ^ key)
   | Some _ ->
       Stdio.printf "\nWelcome to OCANNL! Configuration defaults file is disabled.\n%!";
       Hashtbl.create (module String)
diff --git a/ocannl_config.example b/ocannl_config.example
@@ -11,6 +11,9 @@
 # and in ancestor directories. The source of the `log_level` config is always printed,
 # the sources of other configs are printed when the log level > 0.
 # The configuration values below are (one of) the defaults.
+#
+# Repeating fields are disallowed. If the value is an empty string, the default value
+# is used (as if the config was not present).
 
 # Configurations that are stored as `Utils.settings`:
 
@@ -59,8 +62,8 @@ cuda_printf_fifo_size=
 # The `-O` argument to the compiler executable for the `cc` backend.
 cc_backend_optimization_level=3
 
-# The `cc` backend compiler executable name.
-cc_backend_compiler_command=cc
+# The `cc` backend compiler executable name. By default, `ocamlc -config` field `c_compiler` is used.
+cc_backend_compiler_command=
 
 # The `-O` argument to `gcc` for the `gcc` backend (using gccjit).
 gccjit_backend_optimization_level=3
