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