BrewKeeper · Local acceptance review · 8 November 2026

Buy by mass.
Brew by volume.

A screen-by-screen record of raw-material setup, recipe planning, brew and cellar execution, and the warehouse quantities behind them.

45 unedited browser capturesFirefox · local Rails appIsolated agent test databaseEnglish + Japanese

Refreshed 2026-11-08T03:47:40+00:00: all existing screens were recaptured from the current local application. A new recorded-use screen shows batch history opening inside the material panel while retaining its heading.

500 mLThe recipe and executed dose
0.8 L/kgThe main scenario's fixed conversion
0.625 kgThe warehouse depletion for that dose

The single-input editor stores the factor and its selected basis directly. Both kg/L and L/kg are exercised. The density precision check reloads 1.037123456789 kg/L exactly.

These are fictional local test fixtures. Fixture creation and some event transitions use application services; captions identify service-backed execution. Configuration, legacy-recipe conversion, brewhouse, fermenter, bright-tank and brink forms are exercised in the browser.

Caught and fixed during this review: inventory amendments displayed a 625,000 mg movement as 625,000 mL because the screen reused the brewing source unit. The correction retains source units within the inventory category and otherwise uses the purchasing default. The screen now shows 0.625 kg.

The regression failed before the fix. The browser now submits a 0.5 kg correction and checks that the source becomes 400 mL and remaining stock becomes 1.5 kg. Before the fix · Corrected amendment dialog · Failing regression before the fix

The storyboard

Follow the material through the brewery

Choose a chapter or browse every screen. Click a screenshot to enlarge it; use the arrow keys to move between captures. Each card states the quantities checked against the database.

Jump to test evidence
Verification

What ran, and what passed

Local working tree based on 314549990. QA found and corrected the inventory-amendment unit display. Counts below are separate runs and overlap.

ResultCheck
passedRefreshed captures and current editor
111 tests, 905 assertions; no failures or errors. Includes all eight browser scenarios and 103 raw-material controller checks. All 45 captures are current.
Command and output
mise exec ruby@3.2.8 -- env BK_TEST_DATABASE=agent PARALLEL_WORKERS=1 BK_STORYBOARD_OUTPUT=/Users/paul/src/kbc/brewkeeper/docs/reports/brewing-units-2026-11-08 bin/rails test test/system/production/brewing_units_storyboard_test.rb test/controllers/production/raw_materials_controller_test.rb
Read captured output
passedBrowser acceptance and related regressions
41 tests, 579 assertions; no failures or errors. Eight storyboard scenarios produce 44 captures, including Saved/Reset and click-to-select.
Command and output
mise exec ruby@3.2.8 -- env BK_TEST_DATABASE=agent PARALLEL_WORKERS=1 BK_STORYBOARD_OUTPUT=/Users/paul/src/kbc/brewkeeper/docs/reports/brewing-units-2026-11-08 bin/rails test test/system/production/brewing_units_storyboard_test.rb test/system/production/brewing_units_test.rb test/system/production/raw_material_lot_panel_test.rb test/system/production/japanese_units_test.rb test/system/purchasing/custom_unit_equivalents_test.rb
Read captured output
passedFinal full Minitest suite
5,997 tests, 37,756 assertions; no failures or errors, 36 skips.
Command and output
bin/rails-test-agent test
Read captured output
passedLatest raw-material editor and reference links
103 tests, 707 assertions; no failures or errors. Rerun after two late template/controller-test edits.
Command and output
bin/rails-test-agent test test/controllers/production/raw_materials_controller_test.rb
Read captured output
passedModel, service, controller and component checks
327 tests, 2,066 assertions; no failures or errors.
Command and output
bin/rails-test-agent test test/models/product_brewing_units_test.rb test/services/production_brewing_units_test.rb test/services/yeast_service_test.rb test/controllers/production/raw_materials_controller_test.rb test/models/production/material_resource_plan_test.rb test/models/production/reports/recipe_submission_test.rb test/models/production/reports/production_ledger_test.rb test/models/production/reports/batch_turn_material_totals_test.rb test/controllers/production/inventory_amendments_controller_test.rb test/controllers/production/recipes/costing_controller_test.rb test/components/production/brewing_amount_component_test.rb test/components/units_component_test.rb test/models/warehouse/lot_inventory_location_test.rb
Read captured output
passedPost-fix amendment checks
33 tests, 211 assertions; no failures or errors. New volume-source regression failed before the fix; existing mass source units remain unchanged.
Command and output
bin/rails-test-agent test test/services/production_brewing_units_test.rb test/controllers/production/inventory_amendments_controller_test.rb
Read captured output
passedShared RSpec suite
103 examples, no failures, 2 pending.
Command and output
BK_TEST_DATABASE=agent mise exec ruby@3.2.8 -- bundle exec rspec
Read captured output
passedJavaScript controller checks
14 tests; all passed.
Command and output
node --test test/javascript/units_component_test.mjs test/javascript/production_brewing_amount_test.mjs test/javascript/product_select_test.mjs
Read captured output
passedConversion-basis migration and database constraints
3 tests, 16 assertions; no failures or errors.
Command and output
bin/rails-test-agent test test/migrations/store_brewing_conversion_basis_test.rb
Read captured output
passedJavaScript build
esbuild completed successfully before browser testing.
Command and output
mise exec ruby@3.2.8 -- yarn build
Read captured output
passedStoryboard browser verification
45 images decoded; chapter filtering, enlargement and keyboard navigation passed. No horizontal overflow at desktop or 500 px.
Command and output
mise exec ruby@3.2.8 -- bundle exec ruby docs/reports/brewing-units-2026-11-08/verify_gallery.rb
Read captured output
baselineRuboCop on the changed presenter
Three existing heredoc indentation offences on untouched lines 119, 128 and 140. No offences in the changed method.
Command and output
mise exec ruby@3.2.8 -- bundle exec ruby -rrubocop -e 'exit RuboCop::CLI.new.run(ARGV)' -- --cache false app/presenters/production/inventory_amendment_item.rb --format simple
Read captured output

The final full Minitest suite passed: 5,997 tests, 37,756 assertions, no failures or errors, 36 skips. The initial run had 14 matching baseline failures; these no longer occur after the concurrent working-tree updates. The final browser run also passed all 41 checks after correcting navigation, viewport and retained-filter assumptions in the tests.

Two reference-link/template edits landed during the original full-suite run. The refreshed pass now exercises the current editor, recorded-use navigation and every storyboard scenario: 111 tests and 905 assertions passed. The original full-suite results remain recorded separately below.

Initial run: 14 matching baseline cases
  • SquareOrderSynchronisationTest#test_all_active_locations_import_without_inventory_enrolment_and_new_locations_receive_full_history
  • SquareOutstandingOrdersTest#test_healthy_open_draft_and_completed_orders_rely_on_incremental_polling
  • SquareOutstandingOrdersTest#test_local_inventory_repairs_recover_unresolved_facts_without_another_Square_revision
  • SquareSalesBatchesTest#test_catalogue_failures_are_retried_within_a_batch_without_caching_unavailable_classification
  • SquareSalesBatchesTest#test_catalogue_object_reads_are_shared_across_orders_and_variants_but_separated_by_version_and_batch
  • SquareSalesBatchesTest#test_failed_later_pages_keep_their_cursor_and_retry_without_duplicating_inventory_or_audit_effects
  • SquareSalesBatchesTest#test_repeated_cursors_across_job_boundaries_fail_before_acknowledging_another_page
  • SquareSalesBatchesTest#test_restarts_resume_acknowledged_pages_and_completed_locations_during_a_capture_upgrade
  • SquareSalesBatchesTest#test_sales_jobs_yield_between_pages_and_automatically_queue_their_continuation
  • SquareSalesCaptureTest#test_measured_and_ad_hoc_lines_retain_exact_sales_quantities_and_have_stable_shop_identities
  • SquareSalesRefreshJobTest#test_recovery_enqueues_lost_sales_jobs_and_revocation_works_without_new_scopes
  • TranslationSuggestionServiceTest#test_missing_configuration_and_missing_Discord_acknowledgement_are_failures
  • TranslationSuggestionServiceTest#test_posts_one_structured_suggestion_with_source_page,_locale_and_attribution
  • TranslationSuggestionServiceTest#test_transport_errors_do_not_report_success_or_discard_the_suggestion

Edge cases checked

Screen captures show the user-visible cases. Model, service and migration tests cover the constraints and transaction boundaries that cannot be established by a screenshot.

CaseExpected resultEvidence
Purchasing fallbackBlank brewing settings keep purchasing units; turning volume brewing off clears factor and basis.Frames 01, 07; product model tests
Default factor and unitEnabling volume brewing starts at 1 kg/L. mL is selectable through the shared unit component.Frames 02–04; browser tests
Decimal precision1.037123456789 kg/L is stored and reloaded exactly.Frame 03; product and migration tests
Two explicit baseskg/L multiplies volume to mass; L/kg divides. No reciprocal input is saved.Frames 03, 04, 53, 61; JavaScript and product tests
Zero, negative and non-finite factorsReject zero, negative, NaN and infinity; preserve the last valid product.Frame 06; product and migration tests
Partial settings and unsupported categoriesReject partial pairs, mismatched units and unsupported directions. Volume-purchased material cannot enable the mass/volume toggle.Frame 09; product and database tests
Existing recipesKeep saved kg amounts until the user explicitly converts and saves. Copying also preserves original amounts.Frames 30–32; service and browser tests
Independent added rows and product changesNew rows apply the chosen product default, including switches within one category; saved reload retains units.BrewingUnitsTest; JavaScript tests
Planning-only referencesRecipe and pending batch-linked schedule references warn but permit a valid correction.Frames 13, 15; service tests
Projected inventoryChanging 0.8 to 1 L/kg recalculates projected stock from 1.375 to 1.5 kg without changing actual stock.Frames 14–16; MRP tests
Insufficient and combined demandReject 2,000 mL against 1,600 mL available. Two rows cannot independently spend the same mass stock.Frame 17a; production service test
Stale formExecution reloads a conversion changed after the form was opened; stock and cost use the current valid rate.ProductionBrewingUnitsTest
Brewhouse depletion and cost500 mL at 0.8 L/kg removes 625,000 mg and uses the mass lot cost.Frames 12, 17–18; costing and service tests
Fermenter and bright tankBrowser saves in both phases keep mL doses and deplete kg stock.Frames 40–44; stock and completed-task assertions
Brink executionThe browser records a 500 mL nutrient addition, removes 0.625 kg and keeps yeast volume at 20 L.Frames 50–52; yeast service tests
Purchased yeastStaged pitch does not lock conversion. Knockout consumes 0.625 kg for 500 mL, then locks it.Frames 53–54
Substitution densityUse the substitute factor, 1.5 kg/L; 500 mL removes 0.75 kg. Original stock stays untouched.Frames 61–62
Untracked useNo physical stock movement, but actual untracked usage locks conversion.Frame 60; service tests
Actual-use and source protectionLock factor, basis and source measurement after use. A different default within volume remains permitted.Frames 19–20; product, controller and service tests
Amendment and undoA 0.5 kg correction changes the source to 400 mL. Undo restores stock but historical use keeps conversion locked.Frames 21–23; amendment controller and service tests
Reports and category selectionCosting, recipe-submission projections, actual totals and ledger use inventory equivalents; volume report categories remain supported.Frames 11–12, 24; report tests
Optional conversion flagDisabling volumetric conversion still performs standard internal-unit scaling and does not require a factor.Product conversion tests
Japanese and narrow layoutTranslated labels retain canonical units after selection, save and navigation. Configuration is usable at 500 px.Frames 05, 08; JapaneseUnitsTest; JavaScript tests
Purchasing and stock regressionsCustom purchasing-unit equivalents, lot editing, receipt/deletion behaviour and stock panels retain their existing units.CustomUnitEquivalentsTest; RawMaterialLotPanelTest
Migration upgrade and rollbackUpgrade copies old L/kg values without reciprocals; rollback preserves those values and converts terminating density reciprocals.StoreBrewingConversionBasisTest
Saved conversion, Reset and click-to-selectTyping after a click replaces the whole factor. The saved pair remains visible and Reset restores it without a database write.Frame 03a; browser acceptance assertions
Recorded-use navigationThe lock links to recorded batch use inside the material panel; the heading and warehouse stock remain unchanged.Frame 19a; refreshed browser and controller checks
Scope and limits

This exercises the local working tree, including the uncommitted conversion-basis migration. Tests use barcodes_test_agent and its workers. No customer records were edited in development or production. Capture URLs identify temporary fixtures and will not reopen the same records on the development server.

Service-backed scenarios assert actual stock and event quantities, but do not establish every corresponding browser click path. Automated assertions cover rounding, reports and stock boundaries; screenshots are a visual record, not a substitute for those assertions. The database constraint and migration checks run in a private test schema.