land
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
Feb 15 2024
tidy up
address @tomek's feedback
Feb 14 2024
Appreciate that you took the time to update Test Plan, it looks great.
Feb 13 2024
will squash actually
No longer necessary
Feb 12 2024
No longer necessary
No longer necessary since we aren't updating legacy types to include specialRole.
Didn't read too carefully. Can you make sure removing form doesn't affect functionality?
Seems reasonable, but going to have to be annoying here and request a more fleshed out Test Plan for this.
Can we remove extraneous <>/</>
Looks good, but make sure to address copy feedback before landing.
Looks good, some notes inline. Probably makes sense to create tasks and link before landing instead of addressing right now.
High level looks good, might be worth memoizing some stuff here since some of the search stuff may be "expensive"? Not really sure, and web is generally pretty performant (at least on our machines), so up to you.
Feb 10 2024
Feb 9 2024
No longer necessary after handling things differently for old/current/future clients
update
This diff is no longer relevant because we're not including specialRole for "Legacy" RoleInfos.
update
Looks good!
Seems like straightforward move + flow annotation
consistency
Feb 8 2024
rebase to fix CI before addressing @ginsu's feedback
land
and fix CI..
fix flow
This should be safe to land on its own:
- createInitialRolesForNewThread() is only called by createThread() in thread-creator
- The only fields accessed are newRoles.default.id and newRoles.creator.id, so nothing changes here
unstage extraneous code
update
land
In D10990#317334, @tomek wrote:We will patch the specialRole field back in to minimally encoded RawThreadInfos for future clients (where shouldMinimallyEncodePermissions AND includeSpecialRoleFieldInRoles are true).
Is it something already handled in the stack?
Feb 7 2024
Skimmed, assuming this is straightforward refactor.
Looks like it matches the designs.
Thanks for addressing this feedback. Think it's a lot cleaner that the data is being "transformed" once and "pushed up" so that downstream components/logic don't have to deal with transforming data that isn't "ready to go"