Thanks for contributing! This looks like a reasonable change. Please could you write a test for this to ensure no regressions in future? Probably a comparison test makes most sense. Also, has this been tested with webpack 2? It's due to ship v shortly and so this will need to work with webpack 2 (it probably will)

BTW you only need to generate TypeScript 2.1 output. As part of the move to webpack 2 we're only going to run comparison tests against the latest released version of TypeScript (2.1 at present)

Now webpack 2 has shipped our test pack targets webpack 2 alone. Just merged the test pack changes

@herschel666 - you're probably tearing your hair out right now. Bloody Windows, right? The problem appears to be this

It's expecting this:

chunk    {0} bundle.js (main) 43 bytes [entry] [rendered]
[0] ./.test/issue441/app.ts 43 bytes {0} [built]

But finding this:

chunk    {0} bundle.js (main) 46 bytes [entry] [rendered]
[0] ./.test/issue441/app.ts 46 bytes {0} [built]

Ah - the matter of 3 bytes... such heartache! This is the bane of of our comparison tests; we get these marginal differences which are actually of no matter at all; it fails the test. What I'm planning to do is do a little regex magic on our test pack so we don't actually care about these marginal file size differences in output.txt.

If you want to leave things as they are for now I'll take a look at that (hopefully over the weekend some time but no guarantees) and then we can trigger a re-run and merge. I think you're probably done; I apologise for the pain!

Merged - thanks for the contribution!

This will probably go out with our next release which is likely to be our official webpack 2 release. I've put a placeholder in the changelog accordingly: https://github.com/TypeStrong/ts-loader/blob/master/CHANGELOG.md#v200---not-released-yet

Awesome!