figure out why newer builds of PEG.js cause tests to fail

[michaelficarra]

When switching from pegjs/pegjs@bea6b1f to the newer pegjs/pegjs@fe1ca48, after rebuilding the parser, two tests fail with this output:

$ make test
node_modules/.bin/mocha --compilers coffee:. -u tdd -R dot

  ․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․
  ․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․․

  ✖ 2 of 245 tests failed:

  1) Macros __LINE__:
     AssertionError: 4 === 14
    at global.eq ([redacted]/CoffeeScriptRedux/test/setup.coffee:134:37, <js>:37:12)
    at Context.cache$ ([redacted]/CoffeeScriptRedux/test/macros.coffee:8:6, <js>:4:12)
    at Test.Runnable.run ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runnable.js:200:32)
    at Runner.runTest ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:307:10)
    at Runner.runTests.next ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:353:12)
    at next ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:235:14)
    at Runner.hooks ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:244:7)
    at next ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:192:23)
    at Runner.hook ([redacted]/CoffeeScriptRedux/node_modules/mocha/lib/runner.js:212:5)
    at process.startup.processNextTick.process._tickCallback (node.js:244:9)


  2) Command line execution --eval -i:

  AssertionError: 
    at child.(anonymous function) ([redacted]/CoffeeScriptRedux/test/cli-eval-errors.coffee:24:7, <js>:10:9)
    at ChildProcess.exithandler (child_process.js:540:7)
    at ChildProcess.EventEmitter.emit (events.js:91:17)
    at maybeClose (child_process.js:634:16)
    at Process._handle.onexit (child_process.js:676:5)



make: *** [test] Error 2

[ttilley]

Yeah... complete rewrites tend to do that. The commit comment mentions a fair number of incompatibilities. Without having looked in-depth myself, it does seem that this line might at least be relevant: "all internal identifiers visible to user code in the initializer, actions and predicates are prefixed by |peg$|." Then again, it might also be bugs resulting from an incomplete implementation of pegjs's new <meme>ALL THE BYTECODE</meme> approach.

Is this worth looking at in the near future if I have the time to while teaching myself pegjs, or of low enough priority that even code review for such a change would be worth deferring until after your updated schedule?

[michaelficarra]

Fixed by #318.