-*- outline -*-
+* documentation
+Explain $axiom (and maybe change its name: BTYacc names it goal).
+Complete the glossary (item, axiom, ?).
+
+* report documentation
+Extend with error. The hard part will probably be finding the right
+rule so that a single state does not exhibit to many yet undocumented
+``features''. Maybe an empty action ought to be presented too. Shall
+we try to make a single grammar with all these features, or should we
+have several very small grammars?
+
+* documentation
+Some history of Bison and some bibliography would be most welcome.
+Are there any Texinfo standards for bibliography?
+
* Several %unions
I think this is a pleasant (but useless currently) feature, but in the
future, I want a means to %include other bits of grammars, and _then_
char *sval;
}
-* Experimental report features
-Decide whether they should be enabled, or optional. For instance, on:
-
- input:
- exp
- | input exp
- ;
-
- exp:
- token1 "1"
- | token2 "2"
- | token3 "3"
- ;
-
- token1: token;
- token2: token;
- token3: token;
-
-the traditional Bison reports:
-
- state 0
-
- $axiom -> . input $ (rule 0)
-
- token shift, and go to state 1
-
- input go to state 2
- exp go to state 3
- token1 go to state 4
- token2 go to state 5
- token3 go to state 6
-
- state 1
-
- token1 -> token . (rule 6)
- token2 -> token . (rule 7)
- token3 -> token . (rule 8)
-
- "2" reduce using rule 7 (token2)
- "3" reduce using rule 8 (token3)
- $default reduce using rule 6 (token1)
-
-while with --trace, i.e., when enabling both the display of non-core
-item sets and the display of lookaheads, Bison now displays:
-
- state 0
-
- $axiom -> . input $ (rule 0)
- input -> . exp (rule 1)
- input -> . input exp (rule 2)
- exp -> . token1 "1" (rule 3)
- exp -> . token2 "2" (rule 4)
- exp -> . token3 "3" (rule 5)
- token1 -> . token (rule 6)
- token2 -> . token (rule 7)
- token3 -> . token (rule 8)
-
- token shift, and go to state 1
-
- input go to state 2
- exp go to state 3
- token1 go to state 4
- token2 go to state 5
- token3 go to state 6
-
- state 1
-
- token1 -> token . ["1"] (rule 6)
- token2 -> token . ["2"] (rule 7)
- token3 -> token . ["3"] (rule 8)
-
- "2" reduce using rule 7 (token2)
- "3" reduce using rule 8 (token3)
- $default reduce using rule 6 (token1)
-
-so decide whether this should be an option, or always enabled. I'm in
-favor of making it the default, but maybe we should tune the output to
-distinguish core item sets from non core:
-
- state 0
- Core:
- $axiom -> . input $ (rule 0)
-
- Derived:
- input -> . exp (rule 1)
- input -> . input exp (rule 2)
- exp -> . token1 "1" (rule 3)
- exp -> . token2 "2" (rule 4)
- exp -> . token3 "3" (rule 5)
- token1 -> . token (rule 6)
- token2 -> . token (rule 7)
- token3 -> . token (rule 8)
-
- token shift, and go to state 1
-
- input go to state 2
- exp go to state 3
- token1 go to state 4
- token2 go to state 5
- token3 go to state 6
-
-
-> So, it seems clear that it has to be an additional option :)
-
-Paul:
-
- There will be further such options in the future, so I'd make
- them all operands of the --report option. E.g., you could do
- something like this:
+* --report=conflict-path
+Provide better assistance for understanding the conflicts by providing
+a sample text exhibiting the (LALR) ambiguity.
- --report=state --report=lookahead --report=itemset
- --report=conflict-path
+* report
+Solved conflicts should not be reported in the beginning of the file.
+Rather they should be reported within each state description. Also,
+now that the symbol providing the precedence of a rule is kept, it is
+possible to explain why a conflict was solved this way. E.g., instead
+of
- where "--verbose" is equivalent to "--report=state", and where
- "--report=conflict-path" reports each path to a conflict
- state.
+ Conflict in state 8 between rule 2 and token '+' resolved as reduce.
- (As a minor point, I prefer avoiding plurals in option names.
- It's partly for brevity, and partly to avoid wearing out the
- 's' keys in our keyboards. :-)
+we can (in state 8) report something like
-To implement this, see in the Fileutils the latest versions of
-argmatch and so forth.
+ Conflict between rule 2 and token '+' resolved as reduce
+ because '*' < '+'.
+or something like that.
* Coding system independence
Paul notes:
projects / bison.git / blobdiff