bengl/README.md

## README.md

      
    Raw
  

              README.md
            
          
    It turns out that the node REPL doesn't seem to behave as expected inside a domain. (I'm using node v0.10.22)
Try running repltest.js, which is narrowed down from a much larger REPL-thing I'm working on. You'll get your REPL, but there is some weirdness. If you give it an empty line (i.e. just press Enter/Return), you'll get a syntax error, captured by the domain and ouput via line 6:
> 
SyntaxError: Unexpected token )
    at REPLServer.self.eval (repl.js:112:21)
    at Interface.<anonymous> (repl.js:239:12)
    at Interface.EventEmitter.emit (events.js:95:17)
    at Interface._onLine (readline.js:202:10)
    at Interface._line (readline.js:531:8)
    at Interface._ttyWrite (readline.js:760:14)
    at ReadStream.onkeypress (readline.js:99:10)
    at ReadStream.EventEmitter.emit (events.js:98:17)
    at emitKey (readline.js:1095:12)
    at ReadStream.onData (readline.js:840:14)

You'll get the exact same error if you try to start writing something incomplete, like the start of a function:
> function hello() {
SyntaxError: Unexpected token )
    at REPLServer.self.eval (repl.js:112:21)
    at Interface.<anonymous> (repl.js:239:12)
    at Interface.EventEmitter.emit (events.js:95:17)
    at Interface._onLine (readline.js:202:10)
    at Interface._line (readline.js:531:8)
    at Interface._ttyWrite (readline.js:760:14)
    at ReadStream.onkeypress (readline.js:99:10)
    at ReadStream.EventEmitter.emit (events.js:98:17)
    at emitKey (readline.js:1095:12)
    at ReadStream.onData (readline.js:840:14)

Looking into this, it looks like the REPL's way of dealing with multi-line things, etc. is to capture SyntaxErrors and toss them in a buffer, so as to execute the chunk as a whole once it's complete, i.e. no longer causing a SyntaxError (See repl.js L274-279). Unfortunately, when you wrap a REPL in a domain, it doesn't call this finish function (See repl.js L117-123).
So the result here is that adding the domain short-circuits some of the code that deals with incomplete input. Hopefully I don't have to re-implement parts of the finish function in my error handler.
Any ideas here? Please tell me if I'm just doing something I shouldn't be doing in the first place. The intention is to not crash when async errors happen in the REPL.

  
## repltest.js
var repl = require('repl');

var d = require('domain').create();

d.on('error', function(err) {
    console.error(err.stack || err);
});

d.run(function() {
    repl.start('> ');
});
	var repl = require('repl');

	var d = require('domain').create();

	d.on('error', function(err) {
	console.error(err.stack \|\| err);
	});

	d.run(function() {
	repl.start('> ');
	});