Add an option —use-truffle-regex-with-warn

To print a warning when falling back to Joni because a regular expression can’t be run as a Truffle regex.

Author: Lillian Zhang

src/main/java/org/truffleruby/options/Options.java

@@ -129,6 +129,8 @@ public class Options {
     public final boolean WARN_EXPERIMENTAL;
     /** --use-truffle-regex=false */
     public final boolean USE_TRUFFLE_REGEX;
+    /** --use-truffle-regex-with-warn=false */
+    public final boolean USE_TRUFFLE_REGEX_WITH_WARN;
     /** --argv-globals=false */
     public final boolean ARGV_GLOBALS;
     /** --chomp-loop=false */
@@ -246,6 +248,7 @@ public Options(Env env, OptionValues options, LanguageOptions languageOptions) {
         WARN_DEPRECATED = options.get(OptionsCatalog.WARN_DEPRECATED_KEY);
         WARN_EXPERIMENTAL = options.get(OptionsCatalog.WARN_EXPERIMENTAL_KEY);
         USE_TRUFFLE_REGEX = options.get(OptionsCatalog.USE_TRUFFLE_REGEX_KEY);
+        USE_TRUFFLE_REGEX_WITH_WARN = options.get(OptionsCatalog.USE_TRUFFLE_REGEX_WITH_WARN_KEY);
         ARGV_GLOBALS = options.get(OptionsCatalog.ARGV_GLOBALS_KEY);
         CHOMP_LOOP = options.get(OptionsCatalog.CHOMP_LOOP_KEY);
         GETS_LOOP = options.get(OptionsCatalog.GETS_LOOP_KEY);
@@ -387,6 +390,8 @@ public Object fromDescriptor(OptionDescriptor descriptor) {
                 return WARN_EXPERIMENTAL;
             case "ruby.use-truffle-regex":
                 return USE_TRUFFLE_REGEX;
+            case "ruby.use-truffle-regex-with-warn":
+                return USE_TRUFFLE_REGEX_WITH_WARN;
             case "ruby.argv-globals":
                 return ARGV_GLOBALS;
             case "ruby.chomp-loop":

src/main/ruby/truffleruby/core/truffle/regexp_operations.rb

@@ -64,6 +64,7 @@ def self.match_from(re, str, pos)
     Truffle::Boot.delay do
       COMPARE_ENGINES = Truffle::Boot.get_option('compare-regex-engines')
       USE_TRUFFLE_REGEX = Truffle::Boot.get_option('use-truffle-regex')
+      USE_TRUFFLE_REGEX_WITH_WARN = Truffle::Boot.get_option('use-truffle-regex-with-warn')
 
       if Truffle::Boot.get_option('regexp-instrument-creation') or Truffle::Boot.get_option('regexp-instrument-match')
         at_exit do
@@ -75,7 +76,7 @@ def self.match_from(re, str, pos)
     def self.match_in_region(re, str, from, to, at_start, start)
       if COMPARE_ENGINES
         match_in_region_compare_engines(re, str, from, to, at_start, start)
-      elsif USE_TRUFFLE_REGEX
+      elsif USE_TRUFFLE_REGEX || USE_TRUFFLE_REGEX_WITH_WARN
         match_in_region_tregex(re, str, from, to, at_start, start)
       else
         Primitive.regexp_match_in_region(re, str, from, to, at_start, start)
@@ -115,6 +116,9 @@ def self.match_in_region_tregex(re, str, from, to, at_start, start)
         regex_result = compiled_regex.execBytes(str_bytes, from)
       end
       if bail_out
+        if USE_TRUFFLE_REGEX_WITH_WARN
+          warn "match_in_region_tregex(/#{re}/, \"#{str}\"@#{str.encoding}, #{from}, #{to}, #{at_start}, #{encoding_conversion}, #{start}) can’t be run as a Truffle regexp and fell back to Joni"
+        end
         return Primitive.regexp_match_in_region(re, str, from, to, at_start, start)
       elsif regex_result.isMatch
         starts = []

src/options.yml

@@ -144,6 +144,7 @@ EXPERT:
 
     # Controlling the regular expression engines
     USE_TRUFFLE_REGEX: [use-truffle-regex, boolean, false, 'Use the Truffle regular expression engine for Regexp objects']
+    USE_TRUFFLE_REGEX_WITH_WARN: [use-truffle-regex-with-warn, boolean, false, 'Use the Truffle regular expression engine for Regexp objects and warn when falling back to Joni']
 
 INTERNAL: # Options for debugging the TruffleRuby implementation
   EXPERIMENTAL:

src/shared/java/org/truffleruby/shared/options/OptionsCatalog.java

@@ -83,6 +83,7 @@ public class OptionsCatalog {
     public static final OptionKey<Boolean> WARN_DEPRECATED_KEY = new OptionKey<>(false);
     public static final OptionKey<Boolean> WARN_EXPERIMENTAL_KEY = new OptionKey<>(true);
     public static final OptionKey<Boolean> USE_TRUFFLE_REGEX_KEY = new OptionKey<>(false);
+    public static final OptionKey<Boolean> USE_TRUFFLE_REGEX_WITH_WARN_KEY = new OptionKey<>(false);
     public static final OptionKey<Boolean> ARGV_GLOBALS_KEY = new OptionKey<>(false);
     public static final OptionKey<Boolean> CHOMP_LOOP_KEY = new OptionKey<>(false);
     public static final OptionKey<Boolean> GETS_LOOP_KEY = new OptionKey<>(false);
@@ -596,6 +597,13 @@ public class OptionsCatalog {
             .stability(OptionStability.EXPERIMENTAL)
             .build();
 
+    public static final OptionDescriptor USE_TRUFFLE_REGEX_WITH_WARN = OptionDescriptor
+            .newBuilder(USE_TRUFFLE_REGEX_WITH_WARN_KEY, "ruby.use-truffle-regex-with-warn")
+            .help("Use the Truffle regular expression engine for Regexp objects and warn when falling back to Joni")
+            .category(OptionCategory.EXPERT)
+            .stability(OptionStability.EXPERIMENTAL)
+            .build();
+
     public static final OptionDescriptor ARGV_GLOBALS = OptionDescriptor
             .newBuilder(ARGV_GLOBALS_KEY, "ruby.argv-globals")
             .help("Parse options in script argv into global variables (configured by the -s Ruby option)")
@@ -1221,6 +1229,8 @@ public static OptionDescriptor fromName(String name) {
                 return WARN_EXPERIMENTAL;
             case "ruby.use-truffle-regex":
                 return USE_TRUFFLE_REGEX;
+            case "ruby.use-truffle-regex-with-warn":
+                return USE_TRUFFLE_REGEX_WITH_WARN;
             case "ruby.argv-globals":
                 return ARGV_GLOBALS;
             case "ruby.chomp-loop":
@@ -1433,6 +1443,7 @@ public static OptionDescriptor[] allDescriptors() {
             WARN_DEPRECATED,
             WARN_EXPERIMENTAL,
             USE_TRUFFLE_REGEX,
+            USE_TRUFFLE_REGEX_WITH_WARN,
             ARGV_GLOBALS,
             CHOMP_LOOP,
             GETS_LOOP,