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