Denis Kasak

bfredl is correct. `U+2764` is not a [fully-qualified emoji](http://unicode.org/reports/tr51/#def_fully_qualified_emoji) on its own. It needs to be followed by one of two presentation selectors, either `U+FE0F` to select the emoji variant...

Thanks for triaging and linking to the correct issue. Though apart from Element Web not doing the right thing, this is another failure mode and fallout from the /hierarchy endpoint...

This only makes sense iff: - Device signing is mandated, as in MSC4153, and - Either: - The user was invited into the room via a MSC3917-style mechanism, or -...

I had a short discussion about this with @poljar now. Ideally we would reject all encrypted to-device message types if the sender device is insecure. However, there are some caveats...

I believe that is the complete list (but would like @poljar to confirm) and yes, I would like us to reject cleartext messages unless the type is an exception that...

We don't differentiate enough between the Matrix layer things and the Olm/Megolm layer things. If anything should have the name "Megolm session", it should be the low level packets described...

Related: - https://github.com/matrix-org/matrix-spec-proposals/pull/3414 - https://github.com/matrix-org/matrix-spec-proposals/pull/4362 - https://github.com/element-hq/element-meta/issues/275 - https://github.com/matrix-org/matrix-rust-sdk/issues/5397 - https://github.com/element-hq/element-web/pull/30877

Some more potential problems: - No way to list all threads in a room without crawling the entire room history. This could be mitigated by having the client crawl the...

@Half-shot: > It's important to note here that the ID + text label is used to mainain > consistency of the key without having access to history. E.g. a random...

> [*]: Which makes me think of another problem: is ! disallowed in labels due to it also being used as a separator? More generally, which characters are allowed? On...