1 _____________________________________________________________________________
6 - Next target: TopologicalBag (make it wickedfast: preoptimize)
8 - Forest: keep() and valid() -- can we do this with states
11 - hash Long->long: it's all bogus
13 * huge performance improvement (try for more)
14 * pick back up cleaning up end of Parser.java (Reduction)
15 * some weird edge cases; check last regression test, 'make doc'
18 - Sensible tree-printout
19 - make Tib.Block extend Tree<>
21 - more natural phrasing of metagrammar?
23 - finalize metagrammar and rdp-op's
25 - Deal with the problem of zero-rep productions and whitespace insertion
27 - should Union.add() be there?
28 - should Atom.top() be there?
30 - fix the location stuff, it's broken
31 - decent/better error messages
32 - substring parsing required
38 - RFC2822 (email message/headers)
42 ______________________________________________________________________________
45 - clean up the whole Walk situation
48 - right now I can only lift the last child in a forest... begs
49 the question of what the right representation for Forests is
50 if we need to be able to do lift operations on it.
53 - "Regular Right Part" grammars (NP Chapman, etc)
54 - Attribute unification
56 - serialization of parse tables
57 - inference of rejections for literals
58 - "prefer whitespace higher up" (?)
59 - "ambiguity modulo dropped fragments"?
60 - can this be checked statically?
61 - eliminated statically?
63 ______________________________________________________________________________
66 - Implement a k-token peek buffer (for each state, see if it "dead
67 ends" during the next k Phases based solely on state -- ignoring
70 - Arrange for the SPPF corresponding to dropped subtrees to never be
71 generated (or merged, etc)
73 - Is there any way we can avoid creating a GSS.Node instance for
74 nodes which are transient in the sense that they have only one
77 - Re-read Rekers, particularly the stuff on optimal sharing
79 - Isolate the Element objects from Parse.Table/GSS so we can move
82 - consider allowing a Forest.Body to represent some other Tree whose
83 Body's should be [recursively] considered part of this Forest.
85 - perhaps not: right now we have a nice situation where
86 Forest.Ref instances become immutable once iterator()ed. This
87 also gives us a strong place to to culling with the certainty
88 that we won't throw out a Body which would later be salvaged
89 by some yet-to-be-added dependency.
91 - Figure out if there is a way to:
93 - allow unwrapping of children other than the very last one.
95 - fold repetitions into an array form in Forest, before
96 conversion to Tree. The major problem here is that multiple
97 tree-arrays are possible, all of different lengths. Worse,
98 even if they're all the same length, not all elements belong
99 in the same "possibility vector" as all others. You
100 essentially need a GSS to represent the array, which perhaps
101 is what the unfolded form was in the first place.
103 - Wikipedia grammar (needs to be both lexerless and boolean)
106 => Ordered Choice (";" operator)
108 - bring back in parse-table phase resolution of precedence (just
109 like associativity). This can be inferred from the use of ">"
110 when the rules are in one of these special forms:
121 where "_" is anything and "E" is the defining nonterminal.
122 Essentially what we're looking for is the situation where the
123 leftmost portion of one rule produces another rule, and the
124 rightmost portion of the latter produces the former.
126 I'm not 100% certain that this is as "strong" as the prefer/avoid
127 form (try to prove this, you probably can), but it's "what people
128 intend" most of the time.
130 - implement Johnstone's algorithm for "reduced, resolved LR
131 tables" to eliminate superfluous reductions on
134 ______________________________________________________________________________
137 - Rekers & Koorn note that GLR Substring Parsing can be used to do
138 really elegant and generalized "autocompletion".