Four entities, one shape: `id`, `name`, `chatRoomId` and a per-type kind. So the
convenience function the title asks for is four field reads, and recording that
is most of what this doc does -- the rest is the part that is actually missing.
Nothing in the app can say which subgroup a project's room *is*. A name
pre-filled into the picker gets a room related to its entity by a string
somebody typed once, which is the same bargain a dialect already loses: nothing
in the schema makes a dialect unique, so its identity is `(language, country)`
and never its name. So the link goes on the birth certificate -- the one artefact
in the relationship that is signed, travels, and is already read on both sides --
as an optional `subgroupFor` tag the parent's quorum signs along with the name it
already signs.
Eight phases, one commit each, one migration. Phases 1-3 are a shippable release
on their own and are exactly the cheap version of this feature; 4-7 add the tag,
the `DkgSession` columns that carry it past the ceremony, and the reverse view
that makes it worth having.
Two things Phase 0 found by doing what it says:
A dialect has no detail screen, so it is Phase 2 and not a button on a card. The
first draft said the opposite; the reversal is recorded in the doc rather than
edited out, because the argument against the card is the argument for Phase 7 --
a `ListItem` in a `LazyColumn` has nowhere to put a list of the subgroups made
for an entity, and a dialect is the entity a room is most likely to want several
rooms for.
And the schema number collided. 23, 24 and 25 are taken on an unmerged branch;
this takes 23 anyway, because a number is free relative to the merge base and
pre-allocating 26 moves the renumbering rather than avoiding it.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
