enhanced-resolve doesn't always have /lib/ModuleAliasPlugin.js

[robin-surge]

I currently have a project with enhanced-resolve v3.4.1 and it doesn't have /lib/ModuleAliasPlugin.js; this causes RewirePlugin.js to fail in line 48.

See also: https://github.com/webpack/enhanced-resolve/tree/master/lib

Additionally, in webpack the resolvers are initialised with null in the Compiler constructor and you cannot use apply on null: https://github.com/webpack/webpack/blob/master/lib/Compiler.js#L184

[gempain]

Same problem here. Getting TypeError: Cannot read property 'apply' of null as a consequence of the mentioned problem.

[robin-surge]

For everyone having the same problem: You can use babel and babel-rewire-plugin. It sidesteps the issue (rewire is then handled by babel) and is a nice workaround to get unblocked.

[gempain]

Can you provide an example :)?

[robin-surge]

For what, exactly? Using babel-rewire or for the issue described?

[gempain]

Well, I'm not sure but I guess a karma / wepack config file example :)? It's just so I understand how you use Babel-rewire as a workaround :)

[robin-surge]

Its nothing special about what I did. I just used a different set of frameworks. Instead of going karma -> webpack -> webpack-rewire -> mocha/chai I now use karma -> webpack -> babel -> babel-plugin-rewire -> mocha/chai

You can check out the babel plugin here: https://github.com/speedskater/babel-plugin-rewire

[gempain]

Sweet, thanks for the help ! I'll be looking into it. The only thing I am worried about when using babel is that I've always used webpack, and I am testing a lib that is packed with it. But I guess for unit testing it doesn't really matter how you pack scripts. I am somehow worried that packing something differently in a testing environment and in a production environment could cause issues :/

[robin-surge]

My setup still uses webpack :)

[rensbaardman]

For a solution without babel, you can use my fork rensbaardman/rewire-webpack-plugin, which has webpack 4 support.