getInitialProps is called twice

[haoict]

Hi.
As title above, I created codesandbox. Please take a look.
https://codesandbox.io/s/l2838zpq1q

[duongkimngoc]

same issue here 😢

[devrasec]

sadly I have the same issue

[isaachinman]

So, it seems like this is not related to any Redux HOC, although it appears using the Redux HOC is causing the getInitialProps on the _app.js component to get swallowed or otherwise not run.

I'm still seeing two Index::getInitialProps logs even after removing the withRedux HOC.

When submitting a bug report, it's very helpful to try and narrow it down to the simplest possible reproducible example.

I'll look into this further now. Thanks for your patience everyone.

[isaachinman]

So - getInitialProps is being called twice because you're calling it yourself. You don't need to execute the if (Component.getInitialProps) clause - we're already doing that internally within next-i18next.

I believe this was just a case of misunderstanding and copy/pasting from docs. Let me know if this is clear, and if anyone thinks something should be changed.

[haoict]

@isaachinman
I think the problem is not related to withRedux anymore, but only appWithTranslation.
I removed all redux related things, and still seeing two Index::getInitialProps logs.
Please confirm and reopen this issue

https://codesandbox.io/s/l2838zpq1q

[isaachinman]

So: both Component and WrappedComponent have their getInitialProps function called within our appWithTranslation HOC, and weirdly it seems that both are being called onto the index.js level instead of once on _app.js and once on index.js.

If I explicitly do declare a getInitialProps on the _app.js level that doesn't contain another getInitialProps call, the problem does go away. Check out this sandbox.

My guess is that this has to do with hoisting statics and getInitialProps. I suppose we'll need some way of doing an equality check against WrappedComponent and Component.

[isaachinman]

I created a branch called do-not-hoist-getInitialProps which I think fixes this problem.

However, I'm worried that it's also a regression and will break #11. Can a Redux user please test that branch with your compose / connect HOC usage and let me know if everything works?

[haoict]

I've tried to test do-not-hoist-getInitialProps branch, something is broken (in bootstrap.min.js ???). I leave my screenshot here because I'm not sure about things inside.

(In master branch, above error doesn't happen)

[isaachinman]

@haoict I really doubt that's related to our codebase here, but I'd need a reproducible repo to take a look and debug further.

[isaachinman]

Anyone get a chance to do some testing on that branch?

[hwangar]

I came across this issue in my project and confirmed your branch fixes it

[isaachinman]

I know it fixes it. I'm worried about regression with #11.

[isaachinman]

@capellini This is another issue that needs to be resolved before a prod release. Do you have any Redux/HOC projects to test with?

[capellini]

Do you have any Redux/HOC projects to test with?

@isaachinman, unforunately I don't. But in appWithTranslation, we shouldn't have to call both Component.getInitialProps and WrappedComponent.getInitialProps. Calling just WrappedComponent.getInitialProps should suffice. When you use this branch, do things work as expected?

See https://codesandbox.io/s/74l24m5pnj - seems to work in this use case.

[isaachinman]

@capellini What's this loadGetInitialProps util you're pulling in? I don't think it's safe to use NextJs internals. Can you refactor your approach to both:

1. Not use NextJs internals
2. Retain hoisting of getInitialsProps?

If so, that'd be an ideal solution.

[capellini]

@isaachinman loadGetInitialProps removed. See 2773495. Would this work, or do you foresee problems with this approach? I'll submit a PR so we can continue discussion there.

[haoict]

@capellini I think your PR will not resolve this issue.
I've tested it and found that SSR is not working properly. Could you double check?

[capellini]

Yes, you're right. SSR isn't working...looking more into this.

[capellini]

@haoict I think I've fixed the SSR issue. Could you try again and let me know what you find?

[haoict]

@haoict I think I've fixed the SSR issue. Could you try again and let me know what you find?

@capellini seems good now :D

[isaachinman]

@capellini What was the SSR issue? Are you sure that this is safe to do?

[capellini]

The issue was that, since this line was changed from:

    static async getInitialProps({ Component, ctx, ...initObject }) {

to:

    static async getInitialProps(ctx) {

so ctx.ctx === ctx in the old code. I had not accounted for this, so req was always undefined.

It is no less safe than what we were doing before.