philjackson / twerp Goto Github PK

I'm writing a timezone, date format, date parse library in CoffeeScript. I'm using Twerp for testing. There are tests were I run through every day of the year for 70 years and check that the output of a library function matches cached output from GNU date. This makes for a lot of dots on the command line, and it makes me have to scroll back to see if tests passed.

I'd like to see output that is line at a time. A progress indicator that updates itself on one line for a test and then moves onto the next test when it is complete.

I've made a simple update to Twerp that permits natural language test names. Here is an example of the output, using the custom test runner I've created.

http://cloud.github.com/downloads/bigeasy/twerp/twerp120.png

I'd like to suggest that this modification be added to Twerp. It does not break existing tests or test runners.

Would be nice, if you could add a "styled" runner with more spaces and a bit more color.
For example, just change the following functions in progress.js:

Progress.prototype.onStartClass = function(suite) {
this.suite = suite;
return process.stderr.write("\033[33m\n " + suite + "\n");
};

Progress.prototype.onRunEnd = function(summary) {
process.stderr.write("\n\n");
process.stderr.write(" Time taken: " + summary.time + "\n");
process.stderr.write(" Passed: " + summary.passed + "\n");
return process.stderr.write(" Failed: " + summary.failed + "\n");
};

Thanks,
Joachim

When my twerp tests fail due to an exception, I'm expecting to see a stack trace. Instead, the twerp runner stops. With the tiny runner, using the following example, there is no output.

{TwerpTest} = require "twerp"

class exports.ATest extends TwerpTest
  testSomething: (done) ->
    a.b.c
    done 0

It would be nice to get a stack trace and go on with the tests.

Currently, the twerp program has a --runner parameter that is defined using OptionParser as an enum. The enum is populated by a directory listing of lib/runner. This assertion means that, in order to provide a customized test runner, an end user must place the customized runner in the Twerp library.

It would probably be better to allow the user to specify a runner external to Twerp.

Perhaps the --runner parameter can accept a library path. The module would return a CustomRunner class or a createRunner method.