#1042 mistakenly changed the description of state res in v2--v10 rooms

[DMRobertson]

Link to problem area:

• https://spec.matrix.org/v1.3/rooms/v9/#algorithm versus https://spec.matrix.org/v1.2/rooms/v9/#algorithm , steps 1
• Minor drive-by state res clarifications #1042

Issue

In v1.2 and prior:

Take all power events and any events in their auth chains, recursively, that appear in the full conflicted set and order them by the reverse topological power ordering.

In v1.3:

Select all power events that appear in the full conflicted set. Compute the union of their auth chains, including the power events themselves. Sort the union using the reverse topological power ordering.

(emphasis mine).

These are not equivalent. In v1.2 one only selects events in the full conflicted set. In the wording of v1.3, one can pull in auth events from outside the full conflicted set.

I still think the v1.2 text is confusing (the word recursively, in particular). I think we should either

• revert to v1.2 wording
• find a better way to express v1.2's wording. Perhaps:

  Select the set $X$ of all power events in the full conflicted set. For each such power event $P$, enlarge $X$ by adding the events in the auth chain of $P$ which also belong to the full conflicted set. Order $X$ using the reverse topological power ordering.

---

I don't know how the spec release process would handle this (I'm pretty sure I read somewhere that spec releases are meant to be immutable?). Maybe a v1.3.1?

cc @erikjohnston.

[turt2live]

Spec releases can change but it's a royal pain to do so. There's a point where we arbitrarily decide that a change has missed the patch window and can wait until the next release. For something like this, if we get a solution merged before July 6th (arbitrarily) then I'd be happy to do a patch release. Otherwise, it can hold out until v1.4 - it wouldn't be the first time we've shipped a bug in the spec :p