1 _____________________________________________________________________________
4 - Check if the only remaining stack is lame
5 - write a testcase for this
12 foo.add(y.andnot(x)) ==> this is broken
16 ..................................................
18 - evil problems with: (x y? z /ws)
19 - it gets even more evil than that
20 - basically, follow restrictions are not honored when the element
21 matches against the empty string
23 ______________________________________________________________________________
26 - precedes restrictions ("<-")
28 - MUST HAVE BETTER ERROR MESSAGES
29 - use for developing java15.g
32 - once this is ready, do big announcement
34 - topology no longer needed as an arg to parser?
36 - broader regression testing (for stuff like error messages, etc)
38 - More topology untangling [later]
39 - tib: use the lexer only for indentation increases/decreases
40 - grammar highlighting?
42 - Forest needs a "manual access" API
43 - the unwrap bit in Forest makes it really hard to expose an API for forests
47 ______________________________________________________________________________
50 - finalize metagrammar and rdp-op's
55 - RFC2822 (email message/headers)
56 - clean up the whole Walk situation (?)
59 ______________________________________________________________________________
62 - serialization of parse tables
64 - "ambiguity modulo dropped fragments"?
65 - can this be checked statically?
66 - eliminated statically?
68 - substring parsing for better error messages
71 - "Regular Right Part" grammars (NP Chapman, etc)
72 - Attribute unification
74 - inference of rejections for literals
75 - "prefer whitespace higher up" (?)
77 - Labeled edges on trees (associate a label with each slot in the
78 child array in Forest.Body? might make equality tough) --
79 equivalent to Feature Structures. Colon-labeling.
81 ______________________________________________________________________________
84 - understand and implement the RNGLR "kernel state" optimization.
85 The _Practical Early Parsing_ paper may help.
87 - Partly-Linear-PATR? (O(n^6) unification grammar)
89 - Implement a k-token peek buffer (for each state, see if it "dead
90 ends" during the next k Phases based solely on state -- ignoring
93 - Arrange for the SPPF corresponding to dropped subtrees to never be
94 generated (or merged, etc)
96 - Is there any way we can avoid creating a GSS.Node instance for
97 nodes which are transient in the sense that they have only one
100 - Re-read Rekers, particularly the stuff on optimal sharing
102 - Isolate the Element objects from Parse.Table/GSS so we can move
105 - consider allowing a Forest.Body to represent some other Tree whose
106 Body's should be [recursively] considered part of this Forest.
108 - perhaps not: right now we have a nice situation where
109 Forest.Ref instances become immutable once iterator()ed. This
110 also gives us a strong place to to culling with the certainty
111 that we won't throw out a Body which would later be salvaged
112 by some yet-to-be-added dependency.
114 - Figure out if there is a way to:
116 - allow unwrapping of children other than the very last one.
118 - fold repetitions into an array form in Forest, before
119 conversion to Tree. The major problem here is that multiple
120 tree-arrays are possible, all of different lengths. Worse,
121 even if they're all the same length, not all elements belong
122 in the same "possibility vector" as all others. You
123 essentially need a GSS to represent the array, which perhaps
124 is what the unfolded form was in the first place.
126 - Wikipedia grammar (needs to be both lexerless and boolean)
129 => Ordered Choice (";" operator)
131 - bring back in parse-table phase resolution of precedence (just
132 like associativity). This can be inferred from the use of ">"
133 when the rules are in one of these special forms:
144 where "_" is anything and "E" is the defining nonterminal.
145 Essentially what we're looking for is the situation where the
146 leftmost portion of one rule produces another rule, and the
147 rightmost portion of the latter produces the former.
149 I'm not 100% certain that this is as "strong" as the prefer/avoid
150 form (try to prove this, you probably can), but it's "what people
151 intend" most of the time.
153 - implement Johnstone's algorithm for "reduced, resolved LR
154 tables" to eliminate superfluous reductions on
157 ______________________________________________________________________________
160 - Rekers & Koorn note that GLR Substring Parsing can be used to do
161 really elegant and generalized "autocompletion".
164 ______________________________________________________________________________
167 - Incremental parse table construction
168 - "lazy GLR" and "lazy trees" -> language with first-class CF matching
169 - perhaps linear boolean grammars instead? (linear time, quad space)
170 - Forest parsing => chained parsers
171 - unification parsing, attributes, etc
173 - Take another stab at maximal-match? Nonterminal not-followed-by is
175 - Error recovery based on substring parsing