robrix (Rob Rix)

Added a visitor pattern API to HammerRule.

Extracted strings into their own rule so they can be used by the rules for literals and @import directives (DRY, baby!).

Modernized syntax hither and yon.

Modernized HammerMatch.

Added support for ranges in character classes, e.g. [0-9a-fA-F] for hex chars.

Implemented POSIX character classes in character sets, e.g. [[:alpha:]].

Fixed a bug that was causing ignorable rules’ matched text to be misapplied, e.g. in a literal or character set with a space.

Added POSIX character classes to the list, and corrections to \w and \s.

Removed and ignored TM completions.

Added a read me file as a basic getting started guide.

Documented some further esoterica of the use and implementation of Hammer.

Removed baseURL from HammerParser as it was unnecessarily duplicated.

Simplified and tidied the tests.

Reorganized and tidied the project’s files, and changed its bundle identifier to com.monochromeindustries instead of org.garbagecollective.

Significant reorganization and cleanup of HammerParser.

Added a base URL for relative paths, and made much more of the API persist extant errors.

Wrote clearer tests expressing that imported rules override earlier rules and are themselves overridden by later ones.

Added a note about supporting imports relative to some base path.

Fixed a bug with the parser requiring ignorable characters at the end of the parsed source.

Added import rules to grammars, works like mixins.

Corrected a dumb mistake, basically s/objects/methods/ plus a little clarification.

Added a test to demonstrate that you can call methods—even properties—on a variable without compiler warnings by use of the ClassName<Protocol> * type declaration.

Somehow missed that I had appended > to each of the header lines.

RXAssert and RXAssertFalse can now take an optional message which replaces their own logged message.

Updated RXAssertions.

Simplified error handling by removing error contexts and changing the stack to just an error reference and a stack depth.

HammerParser now maintains an array of parser/builder errors so you don’t have to.

Fixed the double-error issue by adding the error in -lengthOfParseFromCursor:withParser:startedAtCursor: instead of -lengthOfMatchFromCursor:withParser:startedAtCursor:, which gets called a second time after checking for ignorable text if it didn’t match the first time around.

Errors are bubbled up to the delegate.

Simplified the error context stack significantly.

Errors are found locally and bubbled up when lookahead fails.

Changed -description’s return type to NSString * as it should always have been.

Changed the README extension in hopes of getting github to notice it.

Added an error context which is maintained by alternation rules.

Synthesize -characterSet on NSCharacterSetRule.

Worked around the broken -invertedSet implementation in 10.5 (at least on PPC).

Described some algorithms for accurate and performant error handling.

Implemented @eschaton’s fine suggestions for Hammer!

Verified that it does the right thing when e.g. an optional method isn’t implemented.