Add grammar spec option to control default state of debug #8
Labels
Comments
mhhollomon
added a commit
that referenced
this issue
Nov 5, 2019
* Add associativity and precedence You can now specify associativity and precedence for terms and precedence for productions. These are used to resolve shift/reduce and reduce/reduce conflicts in the grammar. This completes issue #20 This commit also resolves issue #15 - moving the analyzer to use error_info. More work needs to be done (globally) to use the nested structure of error_info to its advantage, but this at least makes the analyzer use the infrastructure. * Fix up examples so they at least generate. Many of the examples used older syntax or had things that wouldn't parse. This commit fixes them so that they at least parse and generate code. No work was done to make sure the grammars were in any sense "correct". In particular, the sqlite grammar has some known wrongness baked in. These are marked in the grammar. Along the way, some bugs in the parser and error reporting were also fixed. * Add install target issue #23 The release target is only generated in Release mode. However it can be explicitly requested buy setting the option YALR_ENABLE_INSTALL ~~~ cmake -B build -DYALR_ENABLE_INSTALL=ON ~~~ While at it, add options to control creating tests and examples. * Add verbatim sections Code for issue #21. Changes to the calculator example to make use of the new statement. Make some changes to the parser debug output to make the semantic stack a bit more clear - could use some more work. * Add precedence and assocativity statements Add the new statement types and use then in the calculator. This code also contains a partial implementation of a new termset statement. I choose not to document at this time. * Add @lexeme special type enhancement #27 * Cleanup the state table output enhancement #26 . Add a full list of tokens and productions, along with any associated precedence and associativity. * Add conflict information to the state table output The table generator code now saves off the information needed to show that a conflit happened and if it was resolved. The printer routine now use that information to put out conflict info in the Action section of the State descriptions. * Add support for case insensitive match for regex Partial support for enhancement #3. You can now use `rf:` for folding (case insensitive) and `rm:` for explicitly asking for case matching (case sensitive). This will be needed when support is added for changing the default. * Add support for case folded string compares. Fixed string patterns can now be made case insensitive by using the `@cfold` modifer. This also works for skips and regexes. The `rm:` and `ri:` may be deprecated. It is an error to supply both. * Parse option statement More work for enhancement #3 and #8 Add the infrastructure to parse option statements. They don't currently get picked up by the analyzer. * Add new assert macro Add a special `yassert` macro that takes a message argument. Add a `yfail(mesg)` that acts like `yassert(false, mesg)`. Note that these macros are NOT controlled by NDEBUG. * Option flags for case matching behavior enhancement #3. There are now two option flags to control the case behavior of the lexer. * Document new case options finish out enhancement #3 * Fix bug #29 When creating the error message for the lack of a goal rule, no source was being given. Add the source to the parse_tree so that we can reference it in the analyzer. May need to push it on to the table gen code as well. * Small code cleanings Remove obsolete header, remove debugging code. * Add code.main option Add an option to have yalr append the definition of a simple `main()` to the generated output. See included README or template_genmain.hpp for details. * Add error_list class code improvement #28. Added a class that wraps the list of errors. This gives a place to hand the list walk and the logic to add with out duplicating it or leaking the abstration that the list is actually a std:list. * Update docs * Revamp attention pointer in error messages. Enhancement #24. Underline the offending text. If no text was given output ~^~ centered at the location. * Prep docs for next release
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Currently, debug is turned off by default. allow the following options in the spec to default to on.
It could still be switched off by the driver code.
The text was updated successfully, but these errors were encountered: