forest seems to be completed

I seem to have finished the implementation of forests.  Now it remains
the implementation of the chain-rule machine, of which I have a rough
plan now.

1 files changed, 14 insertions, 1 deletions

diff --git a/grammar/src/lib.rs b/grammar/src/lib.rs
index 627ae6f..297cb66 100644
--- a/grammar/src/lib.rs
+++ b/grammar/src/lib.rs
@@ -462,6 +462,19 @@ impl Grammar {
     // REVIEW: Do we have a better way to record expansion information
     // than to compute the transitive closure?
 
+    // REVIEW: We need a way to eliminate those left-linearly expanded
+    // edges whose labels had already been considered, and we need to
+    // preserve the transition of the `left_p` property at the same
+    // time.
+    //
+    // Maybe we could decide to delete those edges in the
+    // `remove_predicate`?  But we cannot access the states of NFA in
+    // that predicate, in the current design, thus we need to refactor
+    // some codes, it seems: we need a way to "compactify" an NFA, by
+    // a key function, in such a way that if two entries have the same
+    // key (determined by the key function), then only one, determined
+    // by another function, remains in the NFA.
+
     /// A transformer of labels to be fed into
     /// [`closure`][nfa::default::nfa::DefaultNFA::closure], with the
     /// predicate that returns true if and only if the label of the
@@ -483,7 +496,7 @@ impl Grammar {
         }
 
         // Compute if this is from left-linear expansion: it is so if
-        // and only if one if either the edges comes from left-linear
+        // and only if either one of the edges comes from left-linear
         // expansion or we are moving across a non-terminal expansion,
         // that is to say, the source of the second edge is the
         // starting edge of a non-terminal.