1 _____________________________________________________________________________
4 - finalize metagrammar and rdp-op's
6 - Lay down the law on the different kinds of Sequence productions
12 - Deal with the problem of zero-rep productions and whitespace insertion
14 - switch maximal to not-followed-by (~/~)
16 - should Union.add() be there?
17 - should Atom.top() be there?
19 - fix the location stuff, it's broken
20 - decent/better error messages
26 - RFC2822 (email message/headers)
29 ______________________________________________________________________________
32 - clean up the whole Walk situation
34 - cleaner solution to "maximal"?
37 - right now I can only lift the last child in a forest... begs
38 the question of what the right representation for Forests is
39 if we need to be able to do lift operations on it.
42 - "Regular Right Part" grammars (NP Chapman, etc)
43 - Attribute unification
45 - serialization of parse tables
46 - inference of rejections for literals
47 - "prefer whitespace higher up" (?)
48 - "ambiguity modulo dropped fragments"?
49 - can this be checked statically?
50 - eliminated statically?
52 ______________________________________________________________________________
55 - Implement a k-token peek buffer (for each state, see if it "dead
56 ends" during the next k Phases based solely on state -- ignoring
59 - Arrange for the SPPF corresponding to dropped subtrees to never be
60 generated (or merged, etc)
62 - Is there any way we can avoid creating a GSS.Node instance for
63 nodes which are transient in the sense that they have only one
66 - Implement "GLR syntactic predicates" -- the ability to do
67 arbitrary lookahead (ie "followed-by" and "not-followed-by" for
68 arbitrary patterns). This enables generalized longest-match and
69 lets us drop the Maximal hack.
71 - Re-read Rekers, particularly the stuff on optimal sharing
73 - Isolate the Element objects from Parse.Table/GSS so we can move
76 - consider allowing a Forest.Body to represent some other Tree whose
77 Body's should be [recursively] considered part of this Forest.
79 - perhaps not: right now we have a nice situation where
80 Forest.Ref instances become immutable once iterator()ed. This
81 also gives us a strong place to to culling with the certainty
82 that we won't throw out a Body which would later be salvaged
83 by some yet-to-be-added dependency.
85 - Figure out if there is a way to:
87 - allow unwrapping of children other than the very last one.
89 - fold repetitions into an array form in Forest, before
90 conversion to Tree. The major problem here is that multiple
91 tree-arrays are possible, all of different lengths. Worse,
92 even if they're all the same length, not all elements belong
93 in the same "possibility vector" as all others. You
94 essentially need a GSS to represent the array, which perhaps
95 is what the unfolded form was in the first place.
97 - Wikipedia grammar (needs to be both lexerless and boolean)
100 => Ordered Choice (";" operator)
102 - bring back in parse-table phase resolution of precedence (just
103 like associativity). This can be inferred from the use of ">"
104 when the rules are in one of these special forms:
115 where "_" is anything and "E" is the defining nonterminal.
116 Essentially what we're looking for is the situation where the
117 leftmost portion of one rule produces another rule, and the
118 rightmost portion of the latter produces the former.
120 I'm not 100% certain that this is as "strong" as the prefer/avoid
121 form (try to prove this, you probably can), but it's "what people
122 intend" most of the time.
124 - implement Johnstone's algorithm for "reduced, resolved LR
125 tables" to eliminate superfluous reductions on
128 ______________________________________________________________________________
131 - Rekers & Koorn note that GLR Substring Parsing can be used to do
132 really elegant and generalized "autocompletion".