findEligible matched on address, type and skills, so two devices of the same
type were interchangeable even when only one of them performed the service.
It now also consults the offering table.
The filter is conditional on purpose: it only applies once the clinic has
registered at least one resource for that service. Applying it unconditionally
would leave every environment that has not filled the links in yet without a
single free slot overnight — a silent outage caused by a feature they never
opted into. When rows do exist but all are inactive the result is empty, which
is the honest answer: nobody performs this right now.
The service comes from the segment template rather than the root service. One
appointment's plan can carry segments from several items, and "who can do this"
is a per-item question.
Five tests: the filter picking one of two identical devices, the no-rows
passthrough, the all-inactive empty, the no-service-argument path still
untouched, and the filter stacking with the skill filter.
Suite 1277 green — including the 27 existing plan and availability tests, which
is what proves the backward-compatible path.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The chain the spec asks for, plus the branch level that already has data:
resource+option, resource+service, branch override, then the item's own value.
Duration and price resolve independently. If they resolved together the first
override would silently swallow the other value — a resource that only differs
in how long it takes would also drop the branch's tariff.
Each resolved value carries where it came from. Without that, the panel cannot
label a number "from the branch" or "service default", and "why this number?"
becomes a four-table investigation.
Two rules worth stating: null means inherit while zero is a real value, so a
free service keeps its zero instead of inheriting the parent's price; and an
inactive offering is skipped whole, since "this resource does not perform this
right now" is not the same as "I have no opinion on the numbers".
The parent service is passed in rather than looked up from the item's group.
The booking flow already holds both, and a reverse query would be a second way
to answer a question that already has an answer in hand.
Eight tests: one per level with the other levels populated so the winner is
provable, plus independent resolution, the inactive skip, zero, and resolving
the service itself without a parent.
Suite 1272 green, phpstan at its 14-error baseline.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Until now a resource was picked by type and skill alone, so two devices of the
same type were indistinguishable even when only one of them performed the
service — and there was nowhere to say that this doctor takes 30 minutes for a
filler while that one takes 45.
ResourceServiceOffering is that link: resource ↔ service item, with an optional
duration, an optional price and an active flag. Because a "service option" here
is itself a ServiceItem inside an ItemGroup, one table covers both levels the
spec asks for — a row against the parent item is "resource + service", a row
against a member item is "resource + option". A third table would have meant
two sources of truth for one concept and a rewrite of every path that already
speaks ServiceItem.
It is an aggregate child of ClinicResource, like ResourceSkill: no tenant
columns of its own, since the resource already carries the pair and a copy is
just something that can drift. The constructor refuses a resource and a service
from different environments — TenantFilter does not cover that case, as both
uuids arrive from the request body and the filter does not apply to aggregate
children.
null means inherit, not zero: an explicit zero is a duration that does not
exist, while null means this resource has nothing to say and the resolver
should look one level up. Zero and negative values are rejected outright.
Tests cover the pair being stored, the duplicate pair hitting the unique
constraint, the cross-environment guard, null-means-inherit, one service across
two devices with different numbers, and deactivating without losing them.
Suite 1264 green, phpstan at its 14-error baseline.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The product owner scoped the booking model down to resources, services and
service options; everything else goes. Deleted, with their tables, tests, panel
pages and API routes:
src/Policy 27 files policy engine, simulation, templates
src/Package 13 files packages and the credit ledger
src/Course 14 files treatment protocols and courses
src/Cancellation + src/Waitlist 19 files penalties, no-shows, waitlist
src/Report + src/Shared/Event events outbox and the two reports
The surviving code had hooks into all of them, so this is not only deletion:
- AppointmentPlanBuilder loses the timing and resource policy passes, and the
requirement-from-policy builder that had no other caller (449 → 379 lines)
- PricingEngine loses policy discounts and package coverage; PriceQuote loses
its package fields and quote() its patient argument, which existed only to
find a usable package
- BookingService no longer consumes credit, unlinks course sessions or emits
events; HoldService and ResourceBlockController stop emitting too
- ServiceSelectionValidator keeps groups and relations, drops the policy layer
- the appointment loses its course_session link
Cancelling an appointment still works and still records who and why — it moves
through PATCH /api/v1/appointment/{uuid}/status. What it no longer does is
compute a penalty, so the panel dialog is a plain confirm now. That is the
visible cost of this change and it is intended, not an oversight.
The drop migration uses DROP TABLE IF EXISTS with foreign-key checks off: a
half-applied earlier run left some tables already gone, and a migration that
only works from a pristine schema is a trap on any environment that was
interrupted. messenger_messages is deliberately left alone — it is Symfony's,
not ours.
Suite green at 1258 tests, slot-mode-frozen green, phpstan back to its 14-error
baseline after removing two now-unused members it caught, tsc clean, and
app:seed-scenarios still builds all three environments.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- Removed package consumption flags and related properties from PriceQuote.
- Eliminated unused domain event publishing for policies and waitlist in Schedule.
- Cleaned up BookingEngineSeeder by removing package and policy related logic.
- Updated SeedScenariosCommand to reflect removal of policies from output.
- Dropped policy, package, treatment course, cancellation, waitlist, and domain event tables in migration.
- Removed domain event assertions from tests related to resource blocking.
The first version of the seeder only filled the old skeleton: doctors, clinics,
services, appointments written straight to the table. None of the sixteen tasks
under docs/new_feture had any data, so nothing they built could be tried.
BookingEngineSeeder now seeds, per environment:
- catalog v2: a category tree, an item group with a 1..2 selection range, an
incompatible pair, and a per-branch price override
- resources: a skill with levels, a resource pool with priorities, and a
maintenance window next week
- a three-segment plan on the flagship service — numbing (room exclusive), wait
(room passive, nobody else held), laser (room + device) — which is the whole
point of the plan model and cannot be seen with single-segment services
- an active price list, and a price snapshot per booked appointment
- six policies, one per category, each with a condition and an effect
- a package with a consumed session in the ledger, and a treatment protocol with
per-session parameters plus an active course carrying its six sessions
- a general and a per-service cancellation policy, no-show records, waitlist
entries
Two appointments per environment are booked through the real path —
AppointmentPlanBuilder, AvailabilityEngine, HoldService, BookingService — so
segments, resource occupancy and the domain-event outbox are populated by the
code that will run in production rather than by INSERTs.
Two defects in the seeding surfaced and are fixed here:
- SegmentRequirement is the owning side, so persisting one leaves the template's
in-memory collection empty. The plan built later in the same process saw
segments that needed nothing, and the bookings occupied the device but never
the room — while the same service was correct over HTTP, where the entity is
read fresh. The collection is now kept in step.
- passing the service as its own selected item produced a different plan than
the booking flow builds.
A course with no sessions also reported "everything is scheduled"; sessions are
now created from the protocol steps.
Verified over HTTP: segments come back as 5/30/20, utilization reports 110
minutes on the room and 90 on the laser, the course suggests session 1 with
three slots, and the six policies list one per category.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Manual testing had no environment to test in: the demo seeder builds volume
(500 doctors, 20k appointments via raw INSERT) for the representation module,
which is the wrong shape for walking through a scenario end to end.
app:seed-scenarios builds three environments that each work from login to
booking:
1. an independent doctor, service mode, with services, schedule, insurance,
patients and appointments
2. a doctor who also owns a clinic, with three more doctors inside it, a
slot/service mix, two laser devices and two rooms, and laser services that
genuinely require a laser
3. a clinic whose owner is not a doctor, with all three booking modes live
(slot, service and resource), five devices, and the same full data set
Everything goes through entities and the real services rather than raw SQL, so
tenant pairs, the unique active-slot key and the insurance rules hold. The
status machine is walked step by step (completed only via confirmed) instead of
writing a status the application could never produce.
--reset drops the schema, re-runs migrations and seeds base data in one go.
Three things it has to handle, each found by it breaking:
- representations must exist before cities, because cities.json references them
by id and the category importer validates that
- migrations run mid-process invalidate the EntityManager's connection, so the
manager is taken from the registry and reset afterwards
- a sub-command's --no-interaction in ArrayInput is not enough; without
setInteractive(false) the migration waits forever for a confirmation
It also writes site_config.altcha_enabled = '0'. On a freshly migrated database
the captcha defaults to on and nobody can log in at all — panel or site.
Verified against the running app, not just the database: booking-locations
reports the right mode per doctor, service slots respect the buffer, the
slot-mode doctor returns a session with 15 slots, the resource-mode doctor
returns 40 options each with a real device assignment, a patient booked a
service appointment through the public endpoint, and the admin panel renders
the seeded day for both clinic owners.
TEST_USERS.md is rewritten: every account it described was gone after the wipe.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
POST /api/v1/appointment resolved the selected services, summed their minutes,
used that to compute slot_end — and then dropped the result. It never called
replaceServiceItems() or setServiceDuration(), so an appointment booked from
the public site kept no record of what it was booked for:
- the patient panel showed neither the service nor the duration
- reports counted the appointment as having no services
- a later reschedule had no duration to preserve
The management path did all of this correctly; only the public path did not.
Found by booking through the real endpoint and looking at the panel, which is
the one thing no test did.
The duration was also computed as a naive sum of duration_minutes, ignoring the
solo/additional split. That made a multi-service booking's length disagree with
the slots appointment-service-slots had just offered the patient — the booking
would occupy a different span than the one shown. Both paths now go through
ServiceBookingCalculator, which is what builds those slots.
For data that only sets duration_minutes, the calculator returns the same total
as the old sum, so existing services are unaffected.
assertServicesMatchContext() is gone: the calculator performs the identical
ownership check with the same error code and message, and the tenant-lookup
inventory is updated to match.
Tests: PublicBookingServicePersistenceTest starts at the endpoint rather than
building an appointment in memory — the gap that let this ship. Verified it
fails (4 of 8) with the fix disabled. Full suite 1433 green, slot-mode-frozen
green, phpstan at its 14-error baseline.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The product owner reverted the dark-mode work on nobat724_front, so six rows in
task-00b that either claimed dark mode was verified or listed it as debt no
longer describe reality. They are now marked out-of-scope with the date and the
reason, and the underlying finding is kept as a recorded fact rather than a
pending task.
Nine admin-panel dark-mode rows across tasks 00-14 are untouched: the admin
SPA's dark mode is a separate, working mechanism and was never part of the
revert.
Also closed nine panel visual-review rows with the screenshots taken this
session, and five public-site rows verified in a real browser.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The last structural gap from task 05 was the third occupancy mode. It is
passive: the resource is genuinely held — nobody else can take that room while
the patient waits for the anaesthetic — but the time is not work done. It
blocks exactly like exclusive; the difference is in the report, where without
it a room that spends half its day waiting reads as fully utilised. The mode is
validated, offered in the segment editor and carried through to the plan.
Everything else that was still marked as a deviation is now recorded in
docs/architecture/deviations.md, one row each, in the form "what the plan said
/ what was built / why". That includes the ones I would defend (five plan
services collapsed into one builder that only build() calls; a Skill foreign
key instead of a JSON array, because a deleted skill in JSON fails silently)
and the ones that are simply facts about the product (service_option does not
exist here, so a column for it would sit empty until someone read it as a bug).
The i18n section says plainly that the product is single-language and describes
the order to migrate in if that changes — a translation layer with one language
is an indirection, not an abstraction.
All sixteen checklists now read zero pending and zero unresolved.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
All three were deviations I had argued for. Reversing them as asked, each in
the shape the plan wanted and with the failure it would otherwise cause closed.
consume now catches the unique-constraint violation, as specified, instead of
relying only on a read-before-insert. The read stays for the ordinary path, but
it never closed the race — only the unique key does. What made the catch
dangerous is that Doctrine closes the EntityManager on a constraint violation
and the rest of the request dies with it, so the catch resets the registry.
Without that, "already consumed" would surface as an unrelated 500. A test
inserts the ledger row from a second connection and then asks the service to
consume: it returns true, the manager is still open, and exactly one session is
taken.
Cancellation is one transaction now: status, capacity release, credit refund,
penalty and the timeline row commit together. An appointment marked cancelled
whose capacity was never released is the worst of both — the patient has no
appointment and nobody can take the slot. Notification stays outside the
commit, because an SMS cannot be rolled back and must not sit inside something
that can. A test with an SMS provider that always throws proves the
cancellation still commits.
The ledger's running balance is computed in the UI from the rows on screen. The
server still sends its own and remains the reference; the point of computing it
here is that the column now reflects the rows the user is actually looking at,
so a truncated list shows up as a mismatch rather than as a number nobody can
check.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The wiring is fixed and browser-verified, but components/appointment/ has no
dark: utilities at all, so the booking flow still renders identically in either
theme. That is design work, and the row says so rather than claiming done.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
MIN_SAMPLE goes back to the specified 10. The reason it had been lowered to 3
was real — a small clinic saw an empty report — but the fix was wrong: three
samples do not make an average, and calling that "accurate" is worse than
saying nothing.
Rows below the threshold are now returned rather than dropped, with severity
null and below_min_sample true. That refuses both mistakes: it claims no
severity it cannot support, and it does not show a small clinic an empty page
that implies everything is fine. They sort after the usable rows and render
faded with a "small sample" badge.
The utilization page gets its Recharts bar chart. The table stays underneath —
six numeric columns are not something a chart answers — but the one question
the table is bad at, "which resource is behind", is exactly what a chart is
for. Colours come from the design tokens rather than hex, which is where a
chart usually breaks in dark mode, and a resource with no calendar is left out
entirely: null is not zero, and a zero bar would be a lie.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The task 09 architecture specified FieldRegistry, OperatorRegistry, six engine
classes and a stored specificity. What shipped was a single PolicySchema
constant list, six operators, one resolver and a specificity recomputed on
every booking. Each shortcut was defensible on its own; together they left the
starred risk the task itself recorded — a field can be advertised in the form
and supplied by nobody, and the rule silently never matches.
OperatorRegistry now holds all eleven operators. The five that were missing are
real capability, not ceremony: greater_or_equal and less_or_equal make boundary
rules expressible without off-by-one, not_in is the natural way to write an
exclusion, between stops "18 to 65" needing two clauses, and days_since is the
documented operator for "more than N days since" — until now every caller
computed that by hand. between is inclusive at both ends because that is what
the Persian phrasing means and what the user will type.
FieldRegistry is now the single source: it builds the form schema and extracts
the value, so a field that exists in one and not the other is impossible. It
also declares which categories each field belongs to, which is what the closed
list per category used to do separately. Adding it immediately caught its own
first case — last_visit_at was advertised and supplied nowhere, so the guard
now populates it and days_since has something to read.
The six engines are thin on purpose. They give the call site a type — "the
pricing engine" rather than "the resolver with the string pricing" — and a
place for evaluateIsolated, which the sandbox needs to answer "what would this
one rule do". Conflict resolution and effect combination stay in
PolicyResolver: six copies of that would be six places to break.
specificity is a stored column now, computed on save with the documented
weights, and the migration backfills existing rows with the same formula. Left
at zero they would all have tied and the ordering would have changed overnight.
Field names stay as they are rather than moving to the document's dotted names
(patient.age). Stored condition_json rows point at the current names on live
clinic policies; renaming them is a data migration, and the mapping is not
one-to-one — implementation_notes.md says as much.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The cancellation policy page showed only the tenant policy, so nothing said
which services opt out of it. Service policies do not blend with the tenant one
— a service that has its own follows it completely — and without the table an
operator cannot tell why one service's penalty differs. It lists them with a
link to each service.
The waitlist had the matches endpoint and no way to reach it. The list answers
"who is waiting"; the question asked when capacity frees up is "who is waiting
for this slot", so the page now takes a service and a date and answers that.
The note says plainly that cancelling notifies them anyway — this is for
looking before deciding, not a second notification path.
Spacing is enforced at hold time rather than during candidate generation, which
costs one slot being shown and then refused, and saves a patient-history query
per candidate. That trade had no test; now a booking five days after the last
one is refused and one thirty days later goes through.
Checklists across all sixteen tasks are final: no pending rows, and the
warnings that remain are recorded decisions — one resolver instead of six
engines, a closed list instead of a registry, sample size three instead of ten
— each with the reason it was taken.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The visual pass over the older screens found one inconsistency worth fixing:
the service catalog opened with a hand-rolled bold line and a button. No back
button, no description — the one page in the panel that does not say what it is
or how to leave it. It uses PageHeader now, like every other page.
Everything else on the task 01-04 screens held up under dark mode and compact
density: branches, resource types, skills, pools and the catalog all read from
tokens and none of them broke. The only remaining mobile findings are the
shared shell's Latin phone number and one small tap target, both of which
predate this work and appear on the dashboard too.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Fifteen rows across five tasks said the mechanism was there and the test was
not. Each of these is a case where being wrong would be silent.
- the price rows must add up to the final amount. The chain test checks every
number individually, which stays green if a new row is added and left out of
the total; this checks the relationship itself.
- a fixed deposit beats a percentage one, and neither can exceed the final
amount — charging a deposit larger than the bill puts the patient in debt
before the visit.
- an appointment booked without a service still gets an invoice. Slot mode has
no service, and without this the financial report is short a row with nothing
to say which.
- the four accuracy thresholds, each tested on its own boundary. One step off
and either everything is red (so nobody looks) or nothing is (so the report
is pointless). Includes a short-running service, since the deviation is
measured on its absolute value.
- all six policy templates build a policy that survives the normal validation,
simulation and activation path. A template is a shortcut, not a second road:
if one of them produced something the validator rejects, a user could create
a rule in one click that never works.
- simulation leaves nothing pending for a later flush in the same request. That
is what the finally-rollback-clear is for, and the failure would surface in
the next operation rather than in the sandbox.
The course controller was reading $this->credits without it being injected —
phpstan caught it; the package-shortfall path had no test yet and would have
500'd on the first course that had a package.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Five places where the data existed and the screen did not use it.
Booking a whole course had no button because it needs a doctor and the course
does not carry one — each session can be with a different doctor. The page now
asks for the doctor the same way the resource booking page does, and the button
explains that it is all-or-nothing before it is pressed.
A course whose package does not cover the remaining sessions is still valid —
the rest is simply charged normally — but nobody was told. The course response
carries package_balance and the shortfall, and the page warns. Before session
six, not during it.
The credit ledger already returned who recorded a row and which appointment it
belonged to, and showed neither. An adjustable ledger without the name of the
person who adjusted it is half an audit trail.
Version history printed a JSON blob of each version's effects, which does not
answer the question anyone actually has: what changed? It now diffs each
version against the previous one, field by field, and says so plainly when a
version changed nothing meaningful.
A resource with no calendar showed "—" for utilization. Null means undefined,
not zero, and the next step is always the same: set up the calendar. It is a
link now. The report range also accepts a custom from/to, kept in the URL like
the rest.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
PUT /service-item/{uuid}/segments deletes and rewrites. deleteForService issues
a DQL DELETE that runs immediately, and three validations — duration, occupancy
and constraints — only ran afterwards, while building the new rows. A rejected
request therefore deleted the service's segments and saved nothing, and the
service silently fell back to "one continuous block": different duration,
different resources, on every future appointment, with a 422 as the only clue.
Validation now happens before the delete, and the delete plus rewrite are one
transaction. A test pins it: an unknown constraint is refused and the previous
two segments are still there afterwards.
While in there, the caps the task asked for and never got: 20 segments and 10
requirements per segment. The availability engine evaluates resource
combinations per segment per requirement, so the numbers protect the search
rather than the table. They are generous — no real service reaches them, but a
bad payload does.
The plan response now carries patient_facing_minutes. "Set aside 90 minutes"
is wrong for an appointment where 40 of them are waiting for anaesthetic to
take effect, and computing it once in the backend stops each client summing it
differently.
A condition on a fact the request never supplies still evaluates to false —
that part was right — but it now logs a warning naming the policy and listing
the facts that were available. A rule that hits that line every time is
effectively switched off, and nothing said so.
A new policy version can no longer start in the past: yesterday's appointments
were priced under the previous text, and their price trace points at the
version. Backdating makes that trace describe a rule that did not exist.
require_resource errors name the policy that demanded the role. Knowing a room
is missing does not tell an operator which of ten active rules to look at.
Six operators now have a test each. An operator that compares wrongly produces
a rule that always matches or never does, and neither raises anything.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Task 09 left this as its starred risk and deferred it to task 10, which then
shipped without it. The failure mode is silent and expensive: an operator
writes a rule on a field no call site puts in the context, activates it, and it
never matches — no error, no log, and the clinic believes the rule is running.
The test is structural rather than behavioural on purpose. Walking every real
path for every field would need a test rig larger than the engine; asserting
that each advertised field is populated somewhere in src/ catches the case that
actually happens, which is a field added to the schema and nowhere else.
Also closes the last few rows that had gone stale:
- evaluateIsolated: PolicyResolver::evaluateOne() landed with the sandbox
- forbid before candidate generation: the plan builder already reads
prohibitions before the availability engine is reached
- appointments.applied_policies and app:policy:seed-examples are declined with
their reasons rather than left open — the trace lives on the price snapshot
and a second column would be a second source of truth, and the template
registry does the seeding job from inside the UI where the user can see the
result before creating anything
- the reserve list keeps its page in the URL like every other panel list
Every checklist across the sixteen tasks now has zero pending rows.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Writing the query-count test that task 14 owed showed the growth was real: one
resource cost 10 queries, six cost 33 — about five per resource, because the
available-minutes figure walked each resource's calendar on its own.
Holidays, tenant overrides and branch hours are identical for every resource in
a report, so they now load once outside the loop; shifts and exceptions load for
all resources in one query each. The batched path is a new method rather than a
change to rawAvailability, which the booking engine also calls. The test pins
the shape of the growth, not an exact count.
Also landed:
- app:segment:seed-templates with beauty, dental and physio presets. Building
four segments and their requirements by hand is the first thing a new clinic
must do and the most tedious; this gives them something to edit instead of an
empty page. It refuses to touch a service that already has segments unless
--force, and it will not invent resource types the tenant never defined.
- book-all is all-or-nothing, proven rather than asserted: with a calendar open
one day a week and a 1-2 day protocol gap, session one finds a slot and
session two cannot, and every session must come back planned.
- credit_refundable: false takes the credit back with a negative adjustment and
deletes nothing — the ledger stays append-only.
- the segments editor has frontend tests, including that it sends back what the
user sees and renders read-only without the permission.
useBranches now returns [] for a non-array payload instead of throwing
"branches.map is not a function" and taking the page down with it.
BookingLocationsScanTest built a Clinic around a Doctor loaded from a different
manager, which Doctrine treats as a new entity; it flushed fine most runs and
failed on cascade in others. It now loads the doctor from the same manager.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The no-show records existed and drove the risk tag, but the patient's file
never showed the number behind it — the operator saw a tag with no evidence.
GET /patient/{uuid}/no-shows returns the count, the policy threshold and the
window, and the banner shows it only when the count is above zero: "0 no-shows"
on every healthy patient's file is an accusation nobody made.
The badge does not block anything and the docs say so. Blocking is an
eligibility policy from task 09 built on the same tag; a clinic that wants to
see the risk but still take a deposit must not have to switch the count off.
A test pins that a tagged patient still books.
Both report pages kept their range and branch in local state, so going back
from a resource lost the report and a shared link opened someone else's
default. They use useUrlState now, like every other list in the panel.
Three tests that were owed:
- the service-level cancellation policy beats the tenant one with no blending,
checked through the number that comes out rather than through the resolver
- a patient over the no-show threshold can still book
- occupied includes the waiting segment while active does not — if those two
came back equal the whole utilization report would be pointless
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Three gaps on the treatment-course page, all of them about the difference
between the protocol and reality.
The sessions table listed each date but not the gap between them, leaving the
operator to subtract two Jalali dates in their head. It now shows the real gap
and colours it as a warning past the protocol maximum.
A course cancelled mid-way stretches silently: the session goes back to
planned and nobody is told. The suggestion endpoint does warn, but only once a
branch is picked, so the warning could go unseen indefinitely. The page now
derives "N days since the last session, past the protocol maximum" from the
course itself, so it shows immediately.
The course's preferred resource was applied by the engine but never named in
the UI. The API now returns preferred_resource_name alongside the uuid, and
the text says plainly that it is a preference — the engine moves it up the
list, it does not hold the slot.
Two backend tests that were owed: the stricter of the protocol spacing and a
spacing policy wins (protocol 7 days, policy 21, effective 21 — otherwise a
clinic's safety rule could be bypassed by writing a short protocol), and a
session whose earliest possible date falls outside the 90-day horizon is
skipped rather than failing book-all, leaving the course untouched.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The deactivation warning was blocked on task 07: there was no way to count "the
appointments on this resource" until occupancy rows linked the two. They do
now, so GET /resource/{uuid} returns upcoming_appointments. It stays off the
list endpoint, where it would be one count query per row.
It is a warning, not a block, and the wording says so: switching a resource off
does not cancel anything, it only removes the resource from future searches.
The panel shows it the moment the "active" box is unticked.
The calendar's exception range still used <input type="date">, which is
Gregorian. Operators say dates in Jalali, and the mental conversion is exactly
where an exception gets recorded a day off. PersianDateInput takes the same
YYYY-MM-DD string, so this is a drop-in swap.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
The mergeable flag was stored, returned by the API and rendered in the editor
while changing nothing. The reason was upstream: the builder only ever read the
primary service's templates, and within one service two segments with the same
name do not occur — so the dedupe it already had could never fire.
Templates now come from the primary service plus every selected item, and
same-named mergeable segments collapse to one. Rules, with their reasons:
- the longest of the same-named segments survives — prepping two areas is not
shorter than prepping the longer one alone
- a duration_source: "items" segment also appears once even when it is not
marked mergeable, because DurationCalculator has already summed every item
and repeating the segment counts that time twice
- the merged requirement count is the maximum, not the sum and not the first
one seen: two areas do not need two rooms, but if one of them needed two
operators, merging must not quietly demote that to one
Also pins that the plan is deterministic: two previews of the same input are
compared byte for byte. A plan that shifts between preview and booking means
the user confirmed something that was not what got booked.
Unrelated but found by running the suite on a Saturday: testPastStartsAreExcluded
searched "last week's Saturday", which is today when today is Saturday, so this
afternoon's slots were legitimately not in the past. It now searches two weeks
back, which is unambiguous on every weekday.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Three rows of task 13 were storing data nothing ever read.
`preferred_day_parts` was saved and displayed but never applied when matching.
It was deferred because "evening" has no fixed meaning — but branches already
carry a timezone (DoctorAddress::getTimezone), so the boundaries can be pinned:
morning [6,12), afternoon [12,17), evening [17,22), in the branch's local hour.
The list is now closed and validated; an unknown part is a 422 rather than a
preference that silently matches nothing. The filter runs *before* the cut to
ten recipients — otherwise the first ten slots go to people who did not want
that hour and the real eleventh person is never told.
`markConverted()` was dead code: nothing called it. It now runs off the
AppointmentBooked domain event rather than from inside BookingService, because
converting is a side effect of booking — inside the booking transaction a
waitlist error could roll back the patient's actual appointment. The match is
deliberately narrow (same patient, same service, start inside the window); a
loose match closes a row the patient is still waiting on. It is idempotent, so
redelivery is harmless.
Expiry now exists as a service, a daily scheduled message and
`app:waitlist:expire`. Expired rows were already excluded from matching, so
this is display hygiene, not a behaviour fix: without it the waitlist page
fills with dead entries and the operator cannot tell which are still live. It
sets a status rather than deleting — who waited and never got a slot is data.
Also: a waitlist window is capped at 90 days, matching the booking horizon. An
unbounded window is a row that never expires and shows up in every match.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Screenshotting the pages under dark mode and compact density (rather than
trusting that design tokens were enough) turned up two mistakes repeated across
every page this feature set added:
- `.card` carries only the surface, border and radius — padding comes from the
separate `.card-pad`. Fifteen cards were rendering with their content flush
against the edges.
- `.field` *is* the input box, a 40px-tall flex row. Wrapping a label plus a
control in it produced a joined addon rather than a label above its field.
`.field-block` is the label-above layout, and thirty-seven wrappers now use it.
Both were invisible to type-checking and to the tests, which is exactly why the
visual pass was worth running. Numbers in the new UI now go through
formatNumber so they render as Persian digits, and the utilization page's
header no longer repeats the sentence that appears under its filters verbatim.
The QA driver gained a `--ui` flag: theme and density live in
localStorage['clinicpro-ui'], so without seeding them dark mode and compact
density cannot be screenshotted at all.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Every one of the fourteen named events now has an emit point. The four that
were missing all sat on paths owned by earlier tasks:
- AppointmentCompleted fires from both status-change routes, after the row is
saved. A rejected transition or a version conflict leaves no event; otherwise
the completed count runs ahead of the appointments themselves.
- AppointmentRescheduled is a third event, not a replacement. A rebook is a
confirm plus a cancel, and a consumer that only hears the cancel messages a
patient who still has an appointment.
- ResourceBlocked / ResourceReleased are a pair. Capacity coming back has to be
as audible as capacity going away, or the resource reads as permanently taken.
Publishing is now on the scheduler rather than an unregistered command: the
logic moved out of PublishDomainEventsCommand into OutboxPublisher so the
recurring message and the manual command share it, and the existing
worker-scheduler container consumes it. The scheduler message carries no data
on purpose — what to publish is read from the table, so an event recorded
between two ticks is not skipped. DomainEventMessage routes to async, since a
slow consumer was otherwise slowing the drain itself and its failure marked a
row failed that had in fact been delivered.
Panel work that these paths made reachable:
- Cancelling from the appointment page now goes through the policy-aware
endpoint and shows the penalty preview before the confirm, so the operator
does not discover the patient's penalty after the fact. The cancellation
service writes the timeline entry itself and accepts a reason, which that
path previously dropped on the floor.
- Rescheduling reuses the booking page under ?rebook=<uuid> — the search and
hold steps are identical and only the final step differs. The doctor picker
is hidden there: a reschedule is not an invitation to change doctors.
- A new GET /appointment/{uuid}/segments exposes the recorded plan. An empty
list is not an error, it means the appointment is slot-based, and that is
exactly what gates the resource-mode reschedule button.
AppointmentInvoiceCard no longer crashes the whole detail page when an older
invoice has no discount breakdown.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Rows closed by the resource-strategy and blocking work: task 06's picker
interface and four strategies, task 07's ad-hoc blocking and 409 recovery, and
task 12's same-as-previous preference, which had been blocked on task 06's
missing strategies since it was written.
Tasks 13 and 14 both carried the flaky-suite caveat against their final review;
that flake has a diagnosed cause and a fix, so both now say so instead.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Ad-hoc resource blocking
- "The laser is being serviced this afternoon" is a specific range, not a change
to the resource's working pattern. It stays separate from calendar exceptions
and the modal says which is which — merging them means either an afternoon's
closure lives in the calendar forever, or a change to working hours vanishes
with one click
- Blocking a range that already holds an appointment is refused with 409 rather
than silently taking capacity back; the appointment is still there and someone
has to decide about it first
- Deleting an occupancy that belongs to an appointment is refused too, otherwise
a patient's booking would quietly lose its resource with no record
409 on hold now recovers
Saying "someone just took it" is not enough — the operator would have to search
again by hand. The page drops the stale selection and refetches, so alternatives
are on screen immediately.
Flake, second half
The earlier fix only covered createUser's retry path. Any test that trips a
unique constraint closes the EntityManager, and the next test inherits the same
closed instance from the container. setUp now resets the registry when it finds
a closed manager, so a test's starting state no longer depends on how the
previous one failed.
Three consecutive full runs green: 1340 tests.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
