- Now there is support for lexer-generation, including unicode-support (UTF-8, UCS-2, UTF-16, UTF-32/UCS-4). I have talked about those features in some previous blog-posts.
- You do not have to use “;;” any longer, just “;” will work, but for compatibility “;;” will be like a single semicolon.
- Fixed a very old overflow when reading the grammar from an input-stream
- With –error-aware-code (the default) error-messages will now work correctly, when you are compiling the generated file and there is an error in your hand-written code, you will get correct references to the corresponding line in the grammar-file, unfortunately kdev-pg-qt’s error-messages are still not perfect (wrong column-numbers etc.).
- Added some command line options, marginal changes in their behaviour
- Changed some messages (spelling etc.)
Maybe, some day, there will be KDev-PG-Qt 1.0 (currently it is called 0.9.5 ;)), so, what is still to do?
- Make it possible to control what should happen when the lexer fails (it should not just be exit(-1))
- Update the documentation at Techbase
- Update the Kate-syntax-file
- Decide which additional unicode-character classes are useful, add some aliases
- Make new command-line-options available in the CMake-macro
- Find out, how to enable C++0x support in CMake allowing different compilers (yes, I am using variadic templates ;))
- Tokens should no longer be declared inside the parser-class (of course it should still be possible for compatibility)
- Forward-declarations for AST-classes
- Integration with QIODevice/KIO (optional)
- Dumping automata (optional)
Of course there could be more nice stuff, but that would not be related to the recent changes, e.g. LL(k)/LL(*)-parsing, look-ahead, possibility to switch rules off, packed AST-nodes, making it usable as a library etc.
I hope there are really no regressions, and the changes will be useful for the small elite of KDevelop-PG-Qt users.