#fail-fast

A failfast module for idiomatic nodejs splitting error handling from successful continuation.

I recognize that it is a matter of style, but I prefer to deal primarily with the success case and let errors either fall into error-handlers or crash the app. That said, I've used code similar to this failfast function in the past to split off errors and decided it was 'bout time I made it its own thing.

I'm tired of seeing:

      if (err) {
        return callback(err);
      }

polluting my boomerang code.

Example

var util = require('util');
var request = require('request');

function timedRequest(uri, callback) {
  var hrstart = process.hrtime();

  request.get(uri,
    function(err, res, body) {
      if (err) {
        return callback(err);
      }
      var hrend = process.hrtime(hrstart);
      util.log(util.format("%ds %dms %d GET %s", hrend[0], hrend[1] / 1000000, res.statusCode, uri));
      callback(null, res, body);
    });
}

... becomes ...

var util = require('util');
var request = require('request');
var failfast = require('fail-fast')

function timedRequest(uri, callback) {
  var hrstart = process.hrtime();

  request.get(uri,
    failfast(callback,
      function(res, body, cb) {
        var hrend = process.hrtime(hrstart);
        util.log(util.format("%ds %dms %d GET %s", hrend[0], hrend[1] / 1000000, res.statusCode, uri));
        cb(null, res, body);
      }));
}

Install

npm install fail-fast

Test

npm test

... or ...

mocha -R spec

To generate code-coverage of tests, run:

mocha --require blanket -R html-cov > coverage.html

You'll find the coverage report in the file coverage.html.

failfast

failfast is a function that takes the following arguments:

• callback - (optional) the callback function that failfast will split error handling for,
• continuation - (optional) a function where successful continuation proceeds,
• thisArg - (optional) an object that will be used as the continuation's this context.

If you call provide no arguments to failfast, it will be benign (although it will still add overhead).

If callback is specified, it is invoked when failfast receives a truthy argument in position 0.

If continuation is specified, it is invoked when failfast receives a falsy argument in position 0. If callback was also specified, it is provided as the last argument to continuation. Any arguments beyond position 0 (the err argument) are forwarded in order to the specified continuation.

If thisArg is specified, it is bound to the this parameter when continuation is invoked.

A note about performance

Any time you extend the call stack you impact performance. An intermediary callback like failfast is no exception. Depending on your scenario the slight overhead may be acceptable - this is something you have to decide for yourself.

failfast is written in a manner that tries to minimize the overhead. Notably, it uses a technique I call switch-dispatch, which is similar to techniques taken by compilers and optimizers; it leverages what we already know, and produces a jump table to avoid copying the array of arguments. switch-dispatch would probably make you puke if you had to write it everywhere, but the performance is sweet.