Claire

It improves consistency but does not actually fundamentally improve things: indeed, unlike what I initially believed, #24835 does not help with situations resulting from `history.replace` calls, which set a `key`...

To actually fix this we need to properly handle `history.replace`. We may want to encode in the history state whether we are allowed to go back in the history, in...

I'm not sure whether this is a good idea, emoji are not in general properly supported in domain names, and I'm not sure this is worth the additional complexity and...

@renchap do you happen to know how we could cleanly hook into the right places to add a bit of logic and a bit of state in `router.history.replace` and `router.history.push`...

> Auto completion on the client side already inserts `@test@🌈🌈🌈.st` correctly, which is why this issue came up in the first place (the server will currently treat this as a...

@Signez thanks for the input! I think I addressed the issue for good this time by building upon your `router.tsx`. This works for now, but if possible, it is probably...

> I think you can achieve it with something like this in `types/react_router.d.ts`: That does not seem to work, `History.LocationState` is still considered to be of type `unknown`.

Thank you for your contribution! Some people argued strongly against having links behave differently whether they are clicked normally or not (see https://github.com/mastodon/mastodon/issues/19652#issuecomment-1302920921), and this PR seems to run contradictory...

> I wonder if it wouldn't be best without it? Meaning people could more easily switch between the single and multiple-column interfaces without changing the settings. But this may also...

Thanks for digging into this! Note that this was supposedly fixed in `rails-i18n` 7.0.2, and we are still using 6.0.0, so the PR as is would reintroduce the issue.