Add new configuration options for diffing runs debug settings

lukstafi · lukstafi · commit 451f36eeac30 · 2025-03-20T23:51:09.000+01:00
- Introduced `diff_max_distance_factor` and `debug_entry_id_pairs` in `ocannl_config.example` for enhanced diffing capabilities.
- Updated `get_debug` function in `utils.ml` to utilize the new configuration options.
- Modified `primitive_ops.ml` to include debug runtime settings.

Signed-off-by: lukstafi &lt;lukstafi@users.noreply.github.com&gt;
diff --git a/arrayjit/lib/utils.ml b/arrayjit/lib/utils.ml
@@ -330,6 +330,21 @@ let get_debug name =
   let diff_ignore_pattern =
     str_nonempty ~f:Re.Pcre.re @@ get_global_arg ~default:"" ~arg_name:"diff_ignore_pattern_pcre"
   in
+  let max_distance_factor =
+    str_nonempty ~f:Int.of_string @@ get_global_arg ~default:"" ~arg_name:"diff_max_distance_factor"
+  in
+  let entry_id_pairs =
+    let pairs_str = get_global_arg ~default:"" ~arg_name:"debug_entry_id_pairs" in
+    if String.is_empty pairs_str then []
+    else
+      String.split pairs_str ~on:';'
+      |> List.filter_map ~f:(fun pair_str ->
+             match String.split pair_str ~on:',' with
+             | [ id1; id2 ] ->
+                 Option.try_with (fun () ->
+                     (Int.of_string (String.strip id1), Int.of_string (String.strip id2)))
+             | _ -> None)
+  in
   if flushing then
     Minidebug_runtime.debug_flushing ?filename ~time_tagged ~elapsed_times ~print_entry_ids
       ~verbose_entry_ids ~global_prefix:name ~for_append:false ~log_level:original_log_level ()
@@ -350,7 +365,7 @@ let get_debug name =
              ~hyperlink ~toc_specific_hyperlink:"" ~highlight_terms
              ~exclude_on_path:Re.(str "env")
              ~backend ~log_level:original_log_level ?snapshot_every_sec ?prev_run_file
-             ?diff_ignore_pattern filename
+             ?diff_ignore_pattern ?max_distance_factor ~entry_id_pairs filename
 
 let _get_local_debug_runtime =
   let open Stdlib.Domain in
diff --git a/bin/primitive_ops.ml b/bin/primitive_ops.ml
@@ -10,7 +10,14 @@ module Rand = Arrayjit.Rand.Lib
 
 module type Backend = Arrayjit.Backend_intf.Backend
 
-let graph_t () =
+module Debug_runtime = Utils.Debug_runtime
+
+let _get_local_debug_runtime = Utils._get_local_debug_runtime
+
+[%%global_debug_log_level 9]
+[%%global_debug_log_level_from_env_var "OCANNL_LOG_LEVEL"]
+
+let%debug_sexp graph_t () : unit =
   Tensor.unsafe_reinitialize ();
   Rand.init 0;
   let module Backend = (val Arrayjit.Backends.fresh_backend ()) in
diff --git a/ocannl_config.example b/ocannl_config.example
@@ -142,3 +142,10 @@ prev_run_prefix=
 # For ppx_minidebug: a pcre syntax regular expression pattern that will be removed from messages
 # before comparison when highlighting differences between runs, allowing certain differences to be ignored.
 diff_ignore_pattern_pcre=
+
+# For ppx_minidebug: (minimal) width of the search band for the minimal edit distance algorithm.
+diff_max_distance_factor=
+
+# For ppx_minidebug: semicolon-separated list of comma-separated entry IDs from
+# the previous and current run, that should be considered the same by the diffing algorithm.
+debug_entry_id_pairs=
