You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

83 lines
2.9 KiB

- new test from CL4832
- CS4531
- testcases for error nodes
- did I miss a change to Python.stg/returnScope?
- there are base classes Tree-/ParserRuleReturnScope
- update old and add new examples
- need protections in scopeAttributeRef? CL4426
- testcase for $ID.int CL4413
- need to override Target.encodeIntAsCharEscape? CL4389
- look into buildbot
- link in report mails is broken
- timezone bug in p4 scraper
- core:
- only look at changes in src/ & runtime/Java
- quick
- incremential build
- sanity check
- full
- depend on quick
- full build
- ant test
- targets
- depend on changes on src/ and runtime/XXX
- depend on successful core/quick build
- nightlybuild
- depend on core/full
- somehow check which targets are ok
- TreeWizard:
- raise exception on parse errors
- document it in wiki
- publish runtime on cheeseshop
- better documentation for output=template w/ full examples
- antlr3.main:
- verbose/quiet flag: show/hide warnings
- set options in grammar?
- write optionparser descriptions
- better output for return objects
- st support
- custom grammar options per target
- make Grammar.legalOptions changeable
- first extract language option, load target class
- pass options not known by Grammar to target
- patch for CS4010 "null check for $scope::var now" once action parser is fixed
- rename @members/@init to @classmembers, @instancemembers?
- gunit?
- testcases error handling in tree parsers
- better test coverage for runtime modules
- documentation
- more documentation in docstrings
- tune doxygen output
- doxygen frontpage
- do not use Set* templates for properties for Python target
- gate with sempred {target.usePropertySetTemplates()}?
- special template for empty alternative -> pass
- complete runtime
- tree.DoubleLinkTree
- tree.ParseTree
- tree.UnBufferedTreeNodeStream
- default values in rule arguments?
- turn some methods into attributes
- (String|CommonToken)Stream.index()
- (String|CommonToken)Stream.size() --> __len__
- get rid of getter/setter in generated code
- document differences to java API
- add methods to emulate java API, but mark 'em as deprecated
- using Stream.index as a state for 'error-already-reported' or memoization
will be a problem when the stream is not a linear buffer
- optimizations which should be explored:
- DFA: perhaps zip() the lists into a tuple
(eot, eof, min, max, accept, special, transition)
for each state. checkout potential performance gain.
- StingStream: look into optimizing LA(). Perhaps use LAk instead of LA(k)
and create the attributes when needed.
- Perform some magic to improve dfaStateSwitch.
- in lexer rules:
LA == u'a' or LA == u'b' or LA == u'c'...
-> LA in (u'a', u'b', u'c', ...)
or "LA in self.set_xyz" with set_xyz as a class member
- tweak CodeGenerator.genSetExpr()
- make BaseTree.nil() an attribute? or singleton?
- psycho??
- ...