From: Ulya Trofimovich <skvadrik@gmail.com>
Date: Tue, 27 Sep 2016 14:48:27 +0000 (+0100)
Subject: Don't merge conflicting tags, choose arbitrary configuration.
X-Git-Tag: 1.0~39^2~275
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=2ead99e4f9e2628faf35f424d57e740b47c6d5cd;p=re2c

Don't merge conflicting tags, choose arbitrary configuration.

Merging conflicting tags doesn't make sense: information is lost
anyway.
---

diff --git a/re2c/src/ir/dfa/closure.cc b/re2c/src/ir/dfa/closure.cc
index 540246e2..1cb052c4 100644
--- a/re2c/src/ir/dfa/closure.cc
+++ b/re2c/src/ir/dfa/closure.cc
@@ -7,7 +7,7 @@ namespace re2c
 {
 
 static void closure_one(closure_t &clos, nfa_state_t *n, bool *tags, bool *badtags, size_t ntags);
-static void merge_tags(bool *oldtags, const bool *newtags, bool *badtags, size_t ntags);
+static void check_tags(const bool *oldtags, const bool *newtags, bool *badtags, size_t ntags);
 
 void closure(const closure_t &clos1, closure_t &clos2, bool *tags, bool *badtags, size_t ntags)
 {
@@ -17,15 +17,6 @@ void closure(const closure_t &clos1, closure_t &clos2, bool *tags, bool *badtags
 	}
 }
 
-void merge_tags(bool *oldtags, const bool *newtags,
-	bool *badtags, size_t ntags)
-{
-	for (size_t i = 0; i < ntags; ++i) {
-		badtags[i] |= oldtags[i] ^ newtags[i];
-		oldtags[i] |= newtags[i];
-	}
-}
-
 /* note [epsilon-closures in tagged NFA]
  *
  * DFA state is a set of NFA states.
@@ -72,9 +63,7 @@ void closure_one(closure_t &clos, nfa_state_t *n, bool *tags, bool *badtags, siz
 				memcpy(tagptr, tags, ntags * sizeof(bool));
 				clos.push_back(clos_t(n, tagptr));
 			} else {
-				// it is impossible to reach the same NFA state from
-				// different rules, so no need to mess with masks here
-				merge_tags(c->tagptr, tags, badtags, ntags);
+				check_tags(c->tagptr, tags, badtags, ntags);
 			}
 			break;
 		}
@@ -82,4 +71,19 @@ void closure_one(closure_t &clos, nfa_state_t *n, bool *tags, bool *badtags, siz
 	--n->loop;
 }
 
+/*
+ * Check configurations for possible conflicts.
+ * In case of conflict choose random configuration (e.g. the first one)
+ * and don't merge: merging only makes sense for tags from different
+ * rules, and it is impossible to reach the same NFA state from different
+ * rules (hence no need to mess with masks here).
+ */
+void check_tags(const bool *oldtags, const bool *newtags,
+	bool *badtags, size_t ntags)
+{
+	for (size_t i = 0; i < ntags; ++i) {
+		badtags[i] |= oldtags[i] ^ newtags[i];
+	}
+}
+
 } // namespace re2c
diff --git a/re2c/src/ir/dfa/determinization.cc b/re2c/src/ir/dfa/determinization.cc
index 548ac58c..fe6f12a9 100644
--- a/re2c/src/ir/dfa/determinization.cc
+++ b/re2c/src/ir/dfa/determinization.cc
@@ -50,8 +50,12 @@ static void merge_tags_with_mask(bool *oldtags, const bool *newtags,
 	bool *badtags, size_t ntags)
 {
 	for (size_t i = 0; i < ntags; ++i) {
-		badtags[i] |= oldmask[i] & newmask[i] & (oldtags[i] ^ newtags[i]);
-		oldtags[i] |= newtags[i];
+		const bool bad = oldmask[i] & newmask[i] & (oldtags[i] ^ newtags[i]);
+		// don't merge conflicting tags, only note the conflict
+		if (!bad) {
+			oldtags[i] |= newtags[i];
+		}
+		badtags[i] |= bad;
 		oldmask[i] |= newmask[i];
 	}
 }