Add lint rule ensuring that a publishing date is valid

[mcollina]

#7838 had a publish date in the future for a reason - to not be published until that date.

We need better processes for this - plus the website should really not render blog post with a publish date in the future.

[avivkeller]

Typically, if a PR isn’t meant to land until a specific date, the author adds the “on hold” label and clearly states that. I don’t see a strong reason to build logic into the site for such a niche use case.

When was the last time we actually needed to delay a blog post from being published immediately?

[bmuenzenmeyer]

IIRC we had some blog posts around the EOL CVEs that were staged, but i could be misremembering.

[ljharb]

Every other blog system I’ve used doesn’t render a published post until the publish date is in the past - that’s how I’d expect it to work, and i wouldn’t consider it a niche use case personally.

[ovflowd]

Every other blog system I’ve used doesn’t render a published post until the publish date is in the past - that’s how I’d expect it to work, and i wouldn’t consider it a niche use case personally.

I agree with the expectation. It's just that on our system if we did that, it'd filter our these blog posts on preview. We could probably filter them on PROD. I also like the idea of ESlint rules.

[avivkeller]

We can add a lint warning that only turns into an error during merge_group?

That way it doesn't fail the PR, but does prevent it from landing

[mcollina]

That would work!

[ovflowd]

We can add a lint warning that only turns into an error during merge_group?

That way it doesn't fail the PR, but does prevent it from landing

I rather fail the PR. And once the date is achieved, just rerun the CI.

[avivkeller]

I don't think we need to fail the PR while it's still in development. A gentle reminder is more effective than a hard block at the PRs early stage. We only need to fail (and block) once it's going to be merged. Re-running the CI is an extra step we can avoid.

[Harkunwar]

Should we just maybe make a bot to leave a comment on the PR that the publish date is a future date. This comment would need to be resolved to merge the PR anyway, so it will function as a warning without breaking our CI.