Add a sectors-style age-range editor to the person form#1872
Conversation
Age ranges previously sat in the generic profile-specific categories block, separated from sectors. Pull the AgeRange type out so it edits as a column next to sectors, and on the profile render the two as side-by- side columns with age ranges as the thinner one — reclaiming the empty space age groups left when it wrapped to its own full-width row. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
| <%= render "shared/sectorable_item_fields", f: sfi, show_admin_flags: true %> | ||
| <% end %> | ||
| <!-- Sectors + Age ranges (side by side; age ranges is the thinner column) --> | ||
| <% age_type, age_cats = (@person_categories_grouped || {}).find { |type, _| type.name == "AgeRange" } %> |
There was a problem hiding this comment.
🤖 From Claude: AgeRange is pulled out of the shared categories loop here and rejected from the other_category_types block below (line 174), so the primary-age hidden field and "mark the primary ones" hint move into this column with it. Other profile-specific types (Workshop Settings) keep their original full-width rendering.
AgeRange isn't a profile_specific category type, so the person edit form never actually showed an age-range editor — age groups could only be set via registration. Add one that mirrors the sector chip UI: add/remove chips with a primary star, but multi-select primaries (a person serves several primary age groups) and no leader flag. On the profile, sectors and age ranges render as two columns; on the form they split 50/50 and wrap as the viewport narrows. Saving the form previously did `categories = submitted ids`, which would wipe every category type the form doesn't show. Scope replacement to the types the form actually edits (via managed_category_type_ids) so a person's non-AgeRange taggings — and their is_primary/legacy_id — survive untouched. Drop category_ids from strong params so assign_attributes no longer pre-wipes categories before assign_associations runs. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
maebeale
changed the title
| if params[key].key?(:managed_category_type_ids) | ||
| # The form only edits certain category types (e.g. age ranges + workshop | ||
| # settings). Preserve taggings of every other type the form never shows so | ||
| # saving can't silently drop them — and assign the union so the join rows |
There was a problem hiding this comment.
🤖 From Claude: This is the core safety fix: the form sends managed_category_type_ids (age ranges + workshop settings), so we only replace those types and union-preserve every other type the form never shows. Because preserved categories stay in the assigned set, their join rows aren't destroyed/recreated — is_primary and legacy_id survive. Org keeps the old full-replace path since it doesn't send the key.
| # profile-specific type shown above. assign_associations preserves taggings | ||
| # of any other type, so saving the form can't drop a person's non-AgeRange | ||
| # category connections (e.g. art types tagged elsewhere). | ||
| @managed_category_type_ids = ([ @age_range_type&.id ] + |
There was a problem hiding this comment.
🤖 From Claude: Note: category_ids was also dropped from person_params. Otherwise #update's assign_attributes(person_params) would full-replace categories= and save before assign_associations runs its scoped/preserving assignment, destroying the non-AgeRange join rows. assign_associations reads category_ids raw from params, so the contract is unchanged.
Make the two boxes equal height by stretching each column and letting the card grow (flex-col + flex-1 + content-start). Replace the age-range add <select> with an "Add age range" button styled and behaving like cocoon's "Add Sector": it inserts a chip containing a picker, the controller keys the chip's primary star to the chosen id and stops a range being added twice, and disables the button once every range is in use. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
| choose(event) { | ||
| const select = event.target | ||
| const chip = select.closest("[data-age-range-picker-target='chip']") | ||
| chip.dataset.categoryId = select.value |
There was a problem hiding this comment.
🤖 From Claude: A new chip uses a select for its id, but the primary star submits person[primary_age_category_ids][] keyed by id — so on select we copy the chosen id onto the star's value. That keeps the multi-primary contract identical to persisted chips (which render the id directly) without changing the controller params.
Match primary_sector: lighting one star clears the others and prepends that chip, so a person has at most one primary age range. Registration's "primary age group" is itself a single-select dropdown, so the data never has more than one primary — the edit UI now reflects that. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
||
| // Single-select primary: lighting one star clears the others and floats that | ||
| // chip to the front, mirroring primary_sector#selectPrimary. | ||
| togglePrimary(event) { |
There was a problem hiding this comment.
🤖 From Claude: Single-select to match sectors. The server (apply_primary_age_groups!) still accepts a set, but registration's primary age group is a single-select dropdown so the data never has >1 primary — the UI now enforces the same. Persisted chips' stars carry the primaryToggle target too, so clearing-others sees them.
Drop the move-to-front on starring a primary age range and render the selected ranges merged in category position order instead of primary-first, so chips don't jump around as you star/unstar. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Mirror the age-range change for sectors: the edit form now renders sector chips alphabetically (sectors have no position column) via a new sectorable_items_ordered, and the primary-sector controller no longer moves the starred chip to the front — so starring doesn't reshuffle them. Profile, recipients, and dashboard views keep leading with the primary via sectorable_items_primary_first. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
| # name (sectors have no position column, so name is the position-equivalent). | ||
| # Unlike sectorable_items_primary_first, the primary is NOT floated to the top, | ||
| # so starring a sector on the form doesn't reshuffle the chips. | ||
| def sectorable_items_ordered |
There was a problem hiding this comment.
🤖 From Claude: Edit-form ordering is intentionally split from display ordering: the form uses this (alpha, primary not floated) so starring doesn't reshuffle chips, while profile/recipients/dashboard keep sectorable_items_primary_first. Sectors have no position column, so name is the position-equivalent — same idea as the age-range chips, which do have positions.
Age ranges now edit through cocoon nested fields like sectors, via a scoped age_range_categorizable_items association with accepts_nested_ attributes_for, instead of the bespoke category_ids/template Stimulus picker. Both pickers share one primary_tag controller (renamed from primary_sector) whose highlight colors come from the Classes API, so the single-primary star behaves identically. Drop the age-range-specific JS controller and partials. Because age ranges no longer ride category_ids, they're removed from managed_category_type_ids — category_ids now carries only workshop settings, and nested attributes manage the age taggings directly. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
| # categorizable_items (AgeRange categories only) so the form's add/remove and | ||
| # primary toggle round-trip as nested attributes — the is_primary flag splits | ||
| # primary vs additional, no separate primary_age_category_ids param needed. | ||
| has_many :age_range_categorizable_items, |
There was a problem hiding this comment.
🤖 From Claude: Scoped association so age ranges edit via cocoon nested attributes (is_primary on the join row is the primary flag — no separate param). Heads-up: AgeGroupTaggable had a private helper of the same name; I renamed it to age_range_items_relation to avoid the concern shadowing this reader (it was being included below the generated association methods, so the private one won).
Two fixes for the cocoon age picker: - age_range_items_ordered now reads the age_range_categorizable_items association the nested attributes build into (not the general categorizable_items), so a range picked before a failed save survives the re-render — its chip and primary star come back. Mirrors how sectors read sectorable_items. - Add a Person at_most_one_primary_age_range validation mirroring SectorsTaggable, so posting two primary age ranges is a form error. Person-only: organizations aggregate several members' primary age groups, so they legitimately have more than one. Update professional_field_identifiers_spec to assert the cocoon chips instead of the old age checkboxes. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
| # editor. Reads the same association the form's nested attributes build into, so | ||
| # unsaved picks survive a failed save (and aren't primary-first — starring | ||
| # shouldn't reshuffle them). Display surfaces lead with the primary instead. | ||
| def age_range_items_ordered |
There was a problem hiding this comment.
🤖 From Claude: The retention fix: read the same age_range_categorizable_items association the form's nested attributes build into. The old version read the general categorizable_items, whose in-memory target doesn't include the freshly-built (unsaved) items after a failed save — so a picked range vanished on re-render. Sectors never had this because sectorable_items is a single association.
|
|
||
| # Count the in-memory set (not a DB query): nested attributes build the items in | ||
| # one transaction, so a row-level check would see none persisted yet. | ||
| def at_most_one_primary_age_range |
There was a problem hiding this comment.
🤖 From Claude: Person-only on purpose (not in the shared concern like sectors): organizations aggregate several affiliated members' primary age groups via tag_age_groups, so an org legitimately has >1 primary. Sectors avoid this because registration tags orgs with primary_ids: [] (orgs never get a primary sector).
1 participant
🤖 PR, suggested 👤 review level: 🔬 Inspect — new editor UI, a category-assignment scoping fix, and a shared cocoon mechanism
Why
The person edit form never had an age-range editor — AgeRange isn't a
profile_specificcategory type, so it was absent from the categories section. Age groups could only be set via registration. This adds an editor, and unifies it with the sector editor.What
Person#age_range_categorizable_itemsassociation +accepts_nested_attributes_for, rendered byshared/_age_range_item_fieldswithlink_to_add_association/link_to_remove_association. Theis_primaryjoin-row flag is the primary marker — no separate param.primary_tagcontroller (renamed fromprimary_sector) drives the single-select primary star for both pickers; highlight colors come from the Stimulus Classes API. Lighting one star clears the others; no reorder — chips keep alphabetical (sectors) / position (age) order. The bespokeage_range_pickercontroller + chip partials are gone.Person#at_most_one_primary_age_rangemirrorsSectorsTaggable— posting two primaries is a form error. Person-only, because orgs aggregate several members' primary age groups.flex-wrap+min-w-[20rem]), stacking as the viewport narrows; on the profile they're two columns with age ranges thinner.TagAssignable): saving used tocategories = submitted_ids, wiping every type the form doesn't show. The form sendsmanaged_category_type_ids(workshop settings only — age ranges go through nested attributes); only those are replaced, so other taggings survive.category_idsdropped from strong params.Tests
spec/requests/people_age_ranges_spec.rb— cocoon render, tag/primary via nested attributes,_destroy, position order, single-primary validation (request + model), retention after a validation error, and non-AgeRange preservation.spec/models/concerns/sectors_taggable_spec.rb—sectorable_items_orderedvssectorable_items_primary_first.spec/requests/events/professional_field_identifiers_spec.rb— updated to assert the cocoon chips instead of the old age checkboxes.