Skip to content

PLAN_SBL_CROSSCHECK.md

Latest commit

 

History

History
95 lines (76 loc) · 6.94 KB

PLAN_SBL_CROSSCHECK.md

File metadata and controls

95 lines (76 loc) · 6.94 KB
Raw

SpeakerBoxLite cross-check — plan

Companion to BACKLOG.md · OTHER_TOOLS.md · existing oracle: packages/ui/test/micka-crosscheck.browser.spec.ts.

Goal

Add speakerboxlite.com as a second external oracle, alongside micka.de, and use it to cross-check every enclosure type it supports that OpenISD also implements. Where SpeakerBoxLite (SBL) supports an enclosure type OpenISD does not yet implement, record that as a backlog gap instead of a test.

Why a second oracle

micka.de's #ideal calculator does not support passive radiators — confirmed by driving its live form: switching to parameterinput mode exposes exactly fs, vas, qts, qms, Re, Le, Rg, ql, rd, qtc, imp_phase, freq_phase, vb2, rd2, lv2, ql2, temp_luft (sealed qtc, vented rd/vb2/rd2/lv2, a leakage-Q knob) — no PR mass, compliance, area, or count fields. A full-text search of the page for "passiv"/"radiator"/"drone" returns zero hits. So micka-crosscheck.browser.spec.ts cannot validate the PR scenario already shipped in the engine (BACKLOG.md — "Shipped").

SpeakerBoxLite's enclosure-type list (confirmed live in its DOM, not from marketing copy) is:

Closed
Vented
4th Order Bandpass
6th Order Bandpass (Parallel)
6th Order Bandpass (Series)
Passive Radiator
Transmission Line

SBL type → OpenISD status → action

SBL enclosure type OpenISD engine support Action
Closed (sealed) Shipped (alignments.ts) Add cross-check scenario + spec case
Vented Shipped Add cross-check scenario + spec case
Passive Radiator Shipped Add cross-check scenario + spec case — this is the main gap micka can't cover
4th Order Bandpass Shipped Add cross-check scenario + spec case
6th Order Bandpass (Parallel) Not implemented — BACKLOG.md P2 "6th-order bandpass (both chambers ported)" No test yet; backlog item already exists but doesn't distinguish parallel/series — refine wording (see below)
6th Order Bandpass (Series) Not implemented Same backlog item — the parallel/series distinction is new information from SBL and isn't currently captured in BACKLOG.md; split into two sub-bullets or note both alignments explicitly
Transmission Line Not implemented — BACKLOG.md P3 "Transmission line / quarter-wave (line length + stuffing)" No test yet; existing backlog item already covers this, no change needed beyond confirming SBL as a future oracle for it

So only the "6th-order bandpass" backlog bullet needs editing (parallel vs. series alignment wasn't a known distinction before this research); transmission line's existing bullet already matches what SBL calls it.

Implementation steps

  1. Extend scenarios.ts — add a box.type: 'bandpass4' variant (bandpass4 has shipped engine support per BACKLOG.md but has no scenario yet, sealed/vented/pr are the only types currently in Scenario['box']['type']) and a pr scenario with expected values. Add a second oracle field alongside micka?: Record<string, string> — e.g. sbl?: Record<string, string> — so a scenario can carry either or both oracles' expected values.

  2. Explore SBL's per-type field names — only the Closed type's fields have been confirmed live so far: Ql (Box losses), Vb(l), Qtc, F3(Hz), plus a VOLUME GENERATION alignment-preset picker (Max flat amplitude Butterworth-B2, Max flat delay Bessel-BL2, Critically-damped, Chebyshev C2, Chebyshev C2 max power — the classical sealed-box alignment names, relevant to the existing BACKLOG.md "Expand vented alignment presets" item, which could be broadened to cover sealed presets too). Vented, 4th Order Bandpass, and Passive Radiator field names are NOT yet confirmed and must be driven live (as this plan's steps 3+ run) before the spec can be written — do not assume field names from the Closed type generalize.

  3. Write packages/ui/test/speakerboxlite-crosscheck.browser.spec.ts, mirroring micka-crosscheck.browser.spec.ts's structure: navigate to SBL, enter driver T/S under "parameterinput"-equivalent entry, select enclosure type, fill box-specific fields, submit/read the results panel, assert against scenario.sbl. Follows the same "external oracle, not normal CI, run on scenario add / formula change" policy as the micka spec's header comment.

  4. Update BACKLOG.md — "Enclosure types & box model" section: split or reword the "6th-order bandpass (both chambers ported)" bullet to note the parallel vs. series tuning distinction SBL exposes as two separate alignments.

  5. Update OTHER_TOOLS.md — add SpeakerBoxLite as a documented external tool (mirroring the existing micka.de entry) and update open question #1 to note micka's confirmed lack of PR support.

Open question for human to decide

Should the Playwright spec drive SBL's actual rendered UI (as the micka spec does — slower, exercises the real form), or is there a lighter-weight path (e.g. an SBL API endpoint observed via network trace) that avoids full page automation? Not investigated — the DOM exploration so far only used document.querySelector / click() via javascript_tool, no network trace was captured.