Unhandled Runtime Error postcss/postcss from 0.3.1 to 8.3.6

[HonkingGoose]

Describe the bug

When comparing postcss/postcss releases from 0.3.1 to 8.3.6, I get a error:

Unhandled Runtime Error

TypeError: mdastNode.children is undefined

Give the steps to reproduce

1. Go to main branch of our code, run yarn
2. yarn start to get the local development server running
3. Go to http://localhost:3000/comparator?repo=postcss%2Fpostcss&from=0.3.1&to=8.3.6
4. See error

What browsers are you seeing the problem on?

Firefox

Have you thought of a possible solution?

No

If you have thought of a solution, please tell us about it!

No response

Do you want to help fix the bug?

No

Is there anything else we need to know?

No response

[Belco90]

Interesting error. I think I have an idea why it happens and might be even fixed in #302. Let's see what happens with this issue after merging that PR!

[HonkingGoose]

I'll try to keep an eye out if this changes things.

Might also be a good "test-case" for a unit/integration test, to verify that this bug is fixed, and remains fixed? If we ever get around to doing those things. 😄

[HonkingGoose]

This is still a problem on current main even after merging #302 which:

The whole unified/rehype/remark ecosystem is upgraded and working fine with Next.js v11.1 now.

[Belco90]

This needs further investigation then.

[sentry-io]

Sentry issue: OCTOCLAIRVOYANT-WEBAPP-4

[Belco90]

This is solved now! After the types improvements I did on #533, where I had to add some extra checks to make sure there weren't type errors, now this comparison doesn't trigger an error anymore.

You can see that here: https://octoclairvoyant.vercel.app/comparator?repo=postcss%2Fpostcss&from=0.3.1&to=8.3.6

Still, the comparison it's a bit weird, but I think that's a different thing. Should we close this one then?

[HonkingGoose]

This is solved now! After the types improvements I did on #533, where I had to add some extra checks to make sure there weren't type errors, now this comparison doesn't trigger an error anymore.

Yay! 🥳

Still, the comparison it's a bit weird, but I think that's a different thing. Should we close this one then?

What do you mean with "a bit weird"?

[Belco90]

There are a few of these groups in the middle of the result:

We can't do nothing about those from now, it seems those releases didn't follow proper sem ver.

[HonkingGoose]

The problem with the groups deserves its own issue, if we want to fix it.

I'll close this issue as we have fixed the unhandled runtime error that this bug report is about. 😄