ScheduledDate is persisted as a midnight calendar value (board create writes
request.ScheduledDate.Value.Date; board patch writes the parsed .Date into a
datetime2 column with no offset). GetTrendAsync treated it as a UTC instant and
converted to America/New_York, so midnight became 19:00/20:00 the previous day
and every board-scheduled work order fell into the prior bucket: a job scheduled
today read as yesterday and overdue, and the Today bucket showed zero.
Bucket by DateOnly.FromDateTime(scheduled.Date) with no conversion, matching
DashboardMetrics and WorkOrderDerivedFields, so Trend and Stats agree on the
same rows. Rewrite the daily and yearly trend tests to assert calendar-date
classification (the removed conversion had encoded the shift into their
expectations) and add a regression test that a midnight-today work order stays
in the Today bucket and is not overdue.
Seed an organization-wide Area catalogue (East, Central, West, California)
with stable ids, add a nullable AreaId to VendorCompany, allow only Admins
to change it through the roster endpoints, expose areas facet metadata and
an areas[n] company-directory filter with the __unassigned__ sentinel.
Stop client writes from changing Locations.AccountId, make the SH-221 migration discoverable, and thread the board-create CancellationToken through lookup and persistence.
SH-44's user story is about not silently orphaning active work. The
confirmation dialog tells the operator, but nothing told the system, so
a deactivation that left work orders open was indistinguishable from one
that had none.
VendorService now takes an ILogger and writes a warning naming the
vendor, the user and the number of work orders left open whenever the
guard is cleared by confirmation. Both the update and delete paths are
covered; nothing is logged when there was nothing to leave open.
SH-44 and SH-82 both left "blocks, or requires explicit confirmation" to
be decided with the team, and the implementation took the blocking
branch. SH-254 settles it the other way: the approved design offers
"Deactivate anyway" beside the list of open work orders.
Deactivation with open work orders is now permitted, but only when the
caller says it has shown them: ConfirmOpenWorkOrders on the update DTO
and a confirmOpenWorkOrders query parameter on the delete route. Absent
the flag the existing guard still throws, so nothing deactivates by
accident and no caller loses the check by omission.
confirmOpenWorkOrders is a required parameter on DeleteVendorAsync
rather than an optional one, so every call site states its intent.
Two review findings on the additive PATCH path:
- AddTechniciansAsync can rename via CompanyFields.Name and write NormalizedName
against the unique index, but the save had no guard. A colliding rename
surfaced as an unhandled 500 from the PATCH action instead of a stable client
conflict. Pre-check the normalized name against other live companies and throw
VendorRosterDuplicateNameException, with a scoped catch around the save for the
race where a competing rename commits in between. The controller maps it to a
409 alongside the existing concurrency conflict.
- Empty-payload validation only rejected a null CompanyFields, so an all-blank
CompanyFields object was forwarded as a company update, bumping RowVersion and
rewriting every technician's LastModificationTime without changing any company
data. Blank fields now collapse to no company change, and a request with
neither technicians nor a real company value fails validation.
