Marius Andra

Haha, that's nifty. Hopefully no browsers will start punishing scripts that do this any time soon. Worth trying for sure.

I think the right call is to expose something like `posthog.flush()`, which would clear all enqueued requests, and return a promise you can await for. As a temporary workaround, the...

Yeah, this is definitely something to do one day! We should have a good [test suite](https://github.com/PostHog/posthog-js/issues/50) before embarking on this journey though.

This is using `react-virtualized`. For reference, here are all the amazing react table libraries I evaluated but didn't land on: - https://github.com/table-library/react-table-library - virtualization flickers, no cell width/height calculation -...

There are still things to improve, but it's looking better and better 

Hmm... how come we need to complicate this? We give each property a timestamp. Then if `$set` comes in, we override if its timestamp is later than the property's timestamp....

I'm still not following. If requests arrive later because of network issues, they should still contain the original timestamp, no? That's why we have all these `now`, `sent_at` and `timestamp`...

Thanks for being patient and taking time to break this down. Now I get the issue and adding this column does make a lot of sense!

Hmm... Well, I'm not sure we should fix *all* HTML errors. Mostly they don't matter, and especially if we override the blockiness of the elements. For example, if we make...

Paginated API list views are such a common pattern throughout the app, that we could benefit from having a shared abstraction between e.g. feature flags, insights, annotations, etc. This could...