Bryce
3ecd115f76
Capture the proven whole-form hx-select swap method as a reusable skill so every later modal migration is cheaper and consistent. No app code changes. - SKILL.md: the per-migration playbook (classify → baseline → characterize → consolidate render fns → templatize → wire HTMX → collapse routes → verify → commit → feed skill) + Growth contract + non-negotiables. - reference/swap-doctrine.md: the four swap rules, focus invariant, Alpine-survives- swap hardening, target-selector strategy — worked from the real edit.clj swaps (memo no-request, account→location targeted cell, amount→totals sibling-tbody, vendor/mode/row whole-form). 0 OOB. - reference/render-functions.md: explicit-data or top-rooted cursor; the MapCursor fake + transaction-account-row-no-cursor* twin as the smell to remove. - reference/form-vs-wizard.md: classification + the data-driven session-backed (formtools SessionStorage) engine that replaces the snapshot round-trip + protocol. - reference/selmer-conventions.md: STUB, validated in Phase 2. - component-cookbook.md / gotchas.md / test-recipes.md / scorecard.md: seeded from what transaction-edit proves (7 cookbook entries, caret-survival + typeahead test recipes, scorecard baseline LOC 1608 / ~12 routes / 1 no-cursor twin / 2 faked roots / 0 OOB). Scorecard (Transaction Edit baseline, before Phase 2): LOC 1608, routes ~12, no-cursor twins 1, faked-cursor roots 2, snapshot merges ~75, OOB 0, mixed hx- 8.
2.6 KiB
Selmer template conventions
Status: STUB — validated in Phase 2. This file describes the target. The Selmer dependency, render helper, and interop bridge are added in Phase 2 (Transaction Edit); rewrite this file from the real, verified example once that lands, and record each converted component in
component-cookbook.md.
Why Selmer for interactive components
In Hiccup the same Alpine/HTMX attribute is sometimes a keyword and sometimes a string in the same file — there's no rule a reader (or an LLM) can rely on:
;; All of these appear in one component today:
:x-ref "input" "x-ref" "hidden"
:x-model "value.value" "x-model" "search"
"@keydown.down.prevent.stop" "tippy.show();" ; handlers MUST be strings
:x-init "..." ; structural attrs are keywords
In a Selmer template the same markup is unambiguous plain HTML:
{# templates/components/typeahead.html #}
<div class="relative" x-data="{{ x_data|safe }}" x-model="{{ x_model }}">
<a class="{{ classes }}" x-ref="input" tabindex="0"
@keydown.down.prevent.stop="tippy?.show()"
@keydown.backspace="tippy?.hide(); value = {value:'', label:''}">
<span x-text="value.label"></span>
</a>
...
</div>
Note the tippy?. null-guard carried over from the swap doctrine — Selmer doesn't change
the Alpine-survives-swap requirement.
Render helper + interop bridge (the Phase 2 foundation)
(defn render [tpl ctx] (selmer/render-file tpl ctx))
(defn hiccup->html [h] (hiccup/html h)) ; embed hiccup inside selmer via {{ frag|safe }}
;; selmer fragment inside hiccup: [:div (hiccup/raw (render "..." ctx))]
The bridge must work both ways during the strangler transition: a Hiccup component
renders inside a Selmer template (pass (hiccup->html h) into the context, render with
|safe), and a Selmer fragment renders inside a Hiccup tree ((hiccup/raw (render ...))).
Prove both in Phase 2 before broad use.
Composition
Selmer composes via {% include %} and {% block %}. Prefer small per-component
templates that the cookbook references by path. Keep |safe to values the server fully
controls (rendered Hiccup, JSON for x-data), never raw user input.
Scope (Open decision 2)
Hybrid: convert interactive/attribute-heavy components first; static markup may stay Hiccup. Revisit a fuller sweep in Phase 11.
Attribute-consistency scorecard (heuristic 8)
grep -cE '"x-[a-z]|"hx-[a-z]|"@' <migrated-template> # → 0 mixed encodings in Selmer
A migrated Selmer template has no mixed :x-/"x-" encodings because everything is plain
HTML.