Essentially, dynamic imports failing on Windows post v2.0.6.
Actual Behavior
I'm opening this issue as I've tested 4.0.0-rc.2 on Windows; and alas the problem with dynamic imports is still present. It's what we use in ts-loader for our execution test pack.
It's nicely illustrated in this PR which upgrades karma from 2-3 (which caused no issues) and then karma-webpack from 2.0.6 to 4.0.0-rc2:
Operating System: Windows 10
Node Version: v10.5.0
NPM Version: v6.1.0
webpack Version: v4.10.2
karma-webpack Version: 4.0.0-rc.2
yarn Version: 1.9.2
First of all, thanks for all the work you do on karma-webpack. It's tremendous; I really appreciate it.
Expected Behavior
This is a reopen of issue #317 which was itself a reopen of #291
The PR https://github.com/webpack-contrib/karma-webpack/pull/338 has not resolved the issue it seems.
Essentially, dynamic imports failing on Windows post v2.0.6.
Actual Behavior
I'm opening this issue as I've tested 4.0.0-rc.2 on Windows; and alas the problem with dynamic imports is still present. It's what we use in ts-loader for our execution test pack.
It's nicely illustrated in this PR which upgrades karma from 2-3 (which caused no issues) and then karma-webpack from 2.0.6 to 4.0.0-rc2:
https://github.com/TypeStrong/ts-loader/pull/840
As you can see, Linux is fine; Windows is not. The failing tests all relate to code splitting / dynamic imports.
How Do We Reproduce?
The simplest way to reproduce this locally is to:
If you'd like to debug this then do this:
And go to http://localhost:9876/
This should run a single test which reliably fails on Windows and passes on Mac / Linux.
To understand the context take a look at the test code here: https://github.com/johnnyreilly/ts-loader/tree/master/test/execution-tests/2.4.1_importCodeSplitting
It should (I hope!) be fairly self-explanatory.