4.0.x broke hot module reloading

[aaronjensen]

On 3.1.1 hot module reloading for stateless functional components worked. On 4.0.0 and 4.0.1 it no longer does for our project.

[sapegin]

I can reproduce it but have no idea why. All HMR configuration should be external.

[aaronjensen]

@sapegin did you change any components from class components to stateless functional components? stateless functional components cannot reload on their own, they must be wrapped at some point by a class component

[sapegin]

I did! Now I’m trying to change them back to classes and it doesn’t help ;-(

[tizmagik]

Hm, I'm not sure if that's it because the root component is a proper class component, so it shouldn't be an issue. I can repro this locally too, not sure what happened. It did reload once but subsequent reloads don't happen, very odd.

[sapegin]

I can reproduce this too in the example. Will try to find the exact commit that breaks it.

[sapegin]

Here is the bad commit: f713a3e. Unfortunately the biggest one. And here I’ve introduced some functional components.

[sapegin]

The only solution I could find is this but I’m not sure it’s better than yours ;-| It’s hack too.

@ Examples.js:15 @
const Examples = ({ examples }) => {
                         <Playground
                             code={example.content}
                             evalInContext={example.evalInContext}
-                             key={index}
+                             key={Math.random()}
                         />
                     );
                 case 'markdown':

[aaronjensen]

that seems worse, it'll recreate everything anytime anything is rerendered won't it?

[sapegin]

Yeah, I think you’re right. How about 3cf04d1? Experimenting with your branch.

[sapegin]

Actually I like this (vanilla Webpack HMR instead of react-transform-hmr or react-hot-loader) s approach more: no need to configure anything and performance shouldn’t be a big issue because you’d want to develop component in isolated mode anyway.

[aaronjensen]

3cf04d1 looks fine (assuming it works). Which approach are you saying you like more? #200?

[sapegin]

Yep, #200 and 3cf04d1 ;-)

[aaronjensen]

Awesome. I'll close this and my pull out. Thanks!

[sapegin]

Thank you too, the solution is 90% yours!