Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
50 commits
Select commit Hold shift + click to select a range
cfd6621
feat(businessevents): export a published service's AsyncAPI contract
claude Nov 18, 2026
7a50cd5
fix(marketplace): refuse update up front when Studio Pro has the proj…
claude Nov 18, 2026
f27e54f
fix(run): take the app and admin ports from the selected configuration
claude Nov 18, 2026
d862964
fix(run): close the diagnostic session when a run command exits
claude Nov 18, 2026
5721392
fix(rest): preview an OpenAPI query parameter as the import stores it
claude Nov 18, 2026
0535f43
fix(odata): name a colliding navigation association <Parent>_<Navigat…
claude Nov 18, 2026
a767b95
check: refuse a mapping custom handler that does not name exactly its…
claude Nov 18, 2026
8da4ec9
fix(marketplace): compare against a reference with widget definitions…
claude Nov 18, 2026
610047b
Merge origin/main
claude Nov 18, 2026
654fd52
fixtures: write json structure samples as `sample`, not the deprecate…
claude Nov 18, 2026
bf71d92
Merge pull request #2392 from ako/claude/busy-sagan-a0u4w0
ako Nov 18, 2026
e10560b
Merge remote-tracking branch 'origin/main' into claude/issue-triage-p…
claude Nov 18, 2026
f3d05b8
Merge remote-tracking branch 'origin/main' into fix/ce0281-from-catalog
claude Nov 18, 2026
3173c68
test: follow the reference cache keeping widgets/ and the <Parent>_<N…
claude Nov 18, 2026
edb32b2
fix(modelsdk): CDC baseline GUIDs read back empty
claude Nov 18, 2026
8da40b5
Merge pull request #2393 from ako/fix/ce0281-from-catalog
ako Nov 18, 2026
5c904a3
Merge origin/main
claude Nov 18, 2026
a591848
check: judge the values of an enumeration split (CE0079, CE1613, CE0771)
claude Nov 18, 2026
47592c4
Merge pull request #2394 from ako/claude/issue-triage-prioritization-…
ako Nov 18, 2026
4549561
Merge origin/main into fix/ce0773-from-catalog
claude Nov 18, 2026
c152c8c
Merge pull request #2395 from ako/claude/busy-sagan-a0u4w0
ako Nov 18, 2026
1e9bf90
Merge pull request #2396 from ako/fix/ce0773-from-catalog
ako Nov 18, 2026
59350fa
fix(lint): parse_xpath reads true()/false() and reports a partial par…
claude Nov 18, 2026
42425d2
fix(translations): match a source text regardless of CRLF or LF line …
claude Nov 18, 2026
8d247c5
fix(enumeration): a value without a caption takes its name as the cap…
claude Nov 18, 2026
52d6b91
fix(pages): address a legacy grid's control-bar buttons and search fi…
claude Nov 18, 2026
2e6804e
fix(release): publish each nightly under its dated tag, and give nigh…
claude Nov 18, 2026
9baee92
Merge origin/main
claude Nov 18, 2026
1b22ee4
test: keep the #2397 doc comment gofmt-stable
claude Nov 18, 2026
30d0705
Merge pull request #2399 from ako/claude/issue-triage-prioritization-…
ako Nov 18, 2026
6c22c9c
chore: ignore generated widget skill docs in testdata fixtures
claude Nov 18, 2026
fead2d4
docs(proposal): verify gate — lint baseline from day one, discoverabi…
claude Nov 18, 2026
6bb96d1
docs(proposal): verify gate — consistent step numbering
claude Nov 18, 2026
d9b154a
external entities: import a key the service does not allow filtering …
claude Nov 18, 2026
0d6b395
chore: ignore mxcli's working directory in top-level testdata fixtures
claude Nov 18, 2026
8d7d1f6
feat(lint): MPR016 reports entities whose boxes overlap on the domain…
claude Nov 18, 2026
f780b96
Merge pull request #2410 from ako/claude/verify-proposal-update
ako Nov 18, 2026
6cd6d40
Merge remote-tracking branch 'origin/main' into fix/ce6630-from-catalog
claude Nov 18, 2026
78a3cc5
docs(proposal): verify gate — MxBuild staging (unmasked/hidden), opti…
claude Nov 18, 2026
f737069
Merge pull request #2402 from ako/claude/wonderful-edison-yxapt0
ako Nov 18, 2026
df56181
Merge branch 'mendixlabs:main' into main
ako Nov 18, 2026
eb26ad4
Merge pull request #2411 from ako/fix/ce6630-from-catalog
ako Nov 18, 2026
3a1e6c9
feat(lint): MPR017 reports a Boolean guard that can only hold one value
claude Nov 18, 2026
dab667d
feat(docker): parse mx check output into structured findings
claude Nov 18, 2026
f64749e
fix(sql): address a generated connector's tables by their schema
claude Nov 18, 2026
78dc2f1
Merge remote-tracking branch 'origin/main' into claude/wonderful-edis…
claude Nov 18, 2026
f77de06
fix(sql): address a generated connector's tables by their schema (#2419)
ako Nov 18, 2026
207fa34
Merge pull request #2412 from ako/claude/verify-proposal-staging
ako Nov 19, 2026
ea4012d
Merge pull request #2418 from ako/claude/mx-check-findings-parser
ako Nov 19, 2026
752e27f
Merge pull request #2420 from ako/claude/wonderful-edison-yxapt0
ako Nov 19, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "cmd-mxcli", "date": "2026-11-18", "refs": ["ako/mxcli#2367", "mendixlabs/mxcli#1882"], "ce": [], "symptom": "Right after `marketplace install` of Workflow Commons 4.10.0, `marketplace diff` reported 2 snippets as locally changed on the run that built the reference and 8 on every cached run (all with pluggable widgets: Timeline, Pop-up menu); `update` then refused without --force and --save-edits saved them as edits.", "file": "cmd/mxcli/marketplace/refcache.go isModelFile (+widgets/) and refEntryFormat f3; cmd/mxcli/marketplace_postinstall.go repairReference; cmd/mxcli/cmd_marketplace_diff.go referenceFor. Tests: cmd/mxcli/marketplace/refcache_widgets_2367_test.go, cmd/mxcli/marketplace_reference_repair_2367_test.go", "cause": "Two differences between the app and its baseline: (1) the reference cache kept the model and action sources but not widgets/, so a cached reference described pluggable widgets generically (`pluggablewidget '…Timeline' …` with the definition's defaults) while the app described `timeline …`; (2) install runs mx update-widgets / rename-design-properties on the app, never on the reference, so the repair's rewrites of stored props read as edits even on a fresh build.", "insight": "**A baseline has to go through every step the subject went through, and its cache has to keep every input DESCRIBE reads.** The 2-vs-8 split between the first and cached runs was the tell: two causes, one in the build and one in the cache. Measured on 11.12.6 with Workflow Commons 4.10.0: untouched install → fresh reference 0 changed (was 2), cached 0 (was 8); with one real `alter snippet` edit → exactly that snippet."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "cmd-mxcli", "date": "2026-11-18", "refs": ["ako/mxcli#2366", "mendixlabs/mxcli#1881"], "ce": [], "symptom": "With Studio Pro's <project>.mpr.lock present, `marketplace update` started, printed ~387 'refusing to write … Studio Pro has this project open' warnings, and ended with 'The project may be mid-update — … Restore from version control', exit 1, although nothing had changed.", "file": "cmd/mxcli/marketplace/update.go PerformUpdate; cmd/mxcli/cmd_marketplace_update.go runMarketplaceUpdate; modelsdk/mpr/studiopro_lock.go CheckStudioProOpen. Test: cmd/mxcli/marketplace/update_studiopro_open_2366_test.go", "cause": "The Studio Pro guard runs per write (guardWrite), so a multi-unit run met it at the first drop and every write after it; the failure was not marked ErrUpdateNotStarted, so the caller printed its 'mid-update' text.", "insight": "**A per-write guard needs a per-run check for runs that cannot stop cleanly.** The guard is right for exec (one write, one refusal); a marketplace update is a drop followed by a transplant, so the same guard turns into hundreds of refusals and a false 'restore from version control'. Ask once before the run, mark the error 'not started', and do it before the expensive reference builds."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "cmd-mxcli", "date": "2026-11-18", "refs": ["ako/mxcli#2360"], "ce": [], "symptom": "diag loop-report over a ~20 h agent session listed every `run wait` (189), `run restart` (150), `run stop` and `run status` call as UNCLOSED with 0.0s total, and blamed 'a process that vanished: killed', although they exited normally and `run wait` alone accounted for over an hour.", "file": "cmd/mxcli/exit_process.go exitProcess; cmd/mxcli/cmd_run.go, cmd/mxcli/cmd_run_lifecycle.go (os.Exit → exitProcess); cmd/mxcli/diag_loop_report.go help. Test: cmd/mxcli/exit_process_2360_test.go", "cause": "runCLI closes the diagnostic session (session_end with duration and exit code) when the root command returns; the run verbs left through os.Exit inside their Run funcs — on exit code 0 too — so no session_end was written.", "insight": "**Instrumentation that lives on the return path misses every os.Exit, including the successful ones.** The earlier fix (closing on the error path) assumed os.Exit meant failure; the lifecycle verbs use the exit code as their API and exit 0 through os.Exit. Route exits through one helper that closes the session, and keep a source test on the files that must not call os.Exit. Measured after: run status (exit 3) and run stop (exit 0) each write session_end; loop-report shows 0 unclosed. 271 other os.Exit sites in package main remain."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "cmd-mxcli", "date": "2026-11-18", "refs": ["ako/mxcli#2365", "mendixlabs/mxcli#1880"], "ce": [], "symptom": "With configuration Default at HttpPortNumber 8085 / ServerPortNumber 8095, `run --local --configuration Default` listened on 8080/8090; only --app-port/--admin-port moved it.", "file": "cmd/mxcli/runports.go runPorts; cmd/mxcli/constants_resolve.go resolveConstantChain (constantChain.AppPort/AdminPort); cmd/mxcli/cmd_run.go. Test: cmd/mxcli/runports_2365_test.go", "cause": "The run took constants and JVM options from the chosen configuration (chooseConfiguration) but its ports only from the flags, whose 0 meant the docker package's 8080/8090 defaults.", "insight": "**Everything a run takes from a configuration should come from the one choice.** Constants and JVM args already rode on constantChain; the ports were the field left out. Measured on 11.12.6 after the fix: the configuration's 8085/8095 → app HTTP 200 on 8085, admin 200 on 8095, nothing on 8080; flags still override each port on its own."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl-backend", "date": "2026-11-18", "refs": ["ako/mxcli#2382", "mendixlabs/mxcli#1889"], "ce": ["CE1613"], "symptom": "On OIDC SSO 2.3.2's OIDC.Snip_Configuration, `alter snippet … set (Caption: 'Edit X') on actionButton4` (a legacy grid control-bar button) failed with 'widget \"actionButton4\" not found'; describe listed the grid's columns but no buttons, though its comment said buttons were addressable; show widgets omitted them.", "file": "mdl/backend/pagemutator/legacygrid.go controlBarItemKeys (finder, nearest-DS, scope walks in mutator.go); mdl/executor/cmd_pages_describe_legacy_grid.go storedItems (+ caption from CaptionTemplate, action); mdl/catalog/builder_pages.go walkWidget. Tests: *_2382_test.go in pagemutator, executor, catalog", "cause": "Forms$GridControlBar.Items and Forms$SearchBar.Items are STORED as `NewButtons` (modelsdk/gen/keyaudit_test.go lists both); every walk read `Items`. #1459's search-field addressing was tested on a synthetic grid built with `Items`, so it never worked on a real project either.", "insight": "**Build a legacy-shape test fixture from a dumped real unit, not from the metamodel names.** `mxcli bson dump -t snippet -o OIDC.Snip_Configuration` showed ControlBar and SearchBar both holding `NewButtons`; the key audit already listed the mismatch but nothing consulted it when #1459 added the walk. Measured on the real snippet: set Caption and set Action on actionButton4 written; describe lists 6 buttons with caption and action; show widgets lists them (after a catalog rebuild — the catalog cache carries no builder version, so an old cache keeps the old index)."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl/executor", "date": "2026-11-18", "refs": ["ako/mxcli#1258"], "ce": ["CE0079", "CE1613", "CE0771"], "symptom": "An enumeration split `case $S/C when … end case` that left an enumeration value without a branch (CE0079), named a value the enumeration does not have, or the right value in another case (CE1613), or named a value in two branches (CE0771) passed `check` and `exec`; the build rejected each. Only the missing `(empty)` branch (MDL056) and an `else` were refused.", "file": "mdl/executor/validate_enum_split_values.go: checkEnumSplitDuplicateValues (MDL105, microflow and nanoflow validators) and enumSplitValueViolations (MDL106, called from TypeCheckProgram next to the expression hints). Tests: validate_enum_split_values_test.go; fixtures mdl-examples/bug-tests/enum-split-duplicate-values.fail.mdl and enum-split-values.mdl", "cause": "MDL056 was written knowing the other half of CE0079, a value with no branch, 'needs the enumeration's member list' and was left undone (its own comment says so); the member list is exactly what the catalog-backed tier already holds (exprcatalog.Reader.EnumCases / AttributeEnumQN, overlaid with the script's own enumerations by declareScriptTypes), so the deferral had outlived its reason.", "insight": "**A comment that defers a check 'until there is context' is a lead: look for the tier that now has the context before measuring from scratch.** CE0773, the code this started from, never appeared on 11.16.1 for any hand-written split (an `else` is a parse error with an explanatory hint, and the older records were the writer's own NoCase bug), so the useful result was the neighbouring gap. Measured one flow per project (a shared project hides CE0079/CE0771 behind CE1613): every value + (empty) and several values per branch clean; a value left out CE0079 on an attribute subject, a parameter subject and a nanoflow; an unknown value or the wrong case CE1613; a value in two branches CE0771 (one per branch, also when the repeat is inside a multi-value branch). An unknown value is reported alone because the build reports CE1613 first. Needs-the-project rules cannot have a `.fail.mdl` fixture: check-mdl runs `check` with no project and requires a fail fixture to fail, so MDL106's negatives live in Go tests (script enumeration, project enumeration through the fixture's Ticket.Status) plus the isolated measurements; MDL105 needs no enumeration and has the fixture. Sweep: 1,328 repo .mdl files, 0 MDL105 hits; 16 files with an enum split under `check -p`, one MDL106 hit - mdl-examples/bug-tests/1929-ended-branch-decision-one-merge.mdl, whose splits name R and G but not B; applied to a fresh project, mx check reports CE0079 for both flows, so the hit is true and the fixture (about describe round trips) was never built against the build. Control: stubbing both rules fails 11 sub-tests; `check -p` agrees with mx check on all 11 measured cases."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl-executor", "date": "2026-11-18", "refs": ["ako/mxcli#2354"], "ce": ["CE0065"], "symptom": "`create external entities` from an OData v4 contract with Product.Category → Category named the association `Integ.Category_2`, a suffix suggesting a clash nothing in the module explained.", "file": "mdl/executor/cmd_contract.go uniqueAssocName (now takes the parent entity). Test: mdl/executor/external_assoc_name_2354_test.go", "cause": "Associations are named after the navigation property; `Category` collides with the entity Category (CE0065 forbids it), and the only fallback was a numeric suffix.", "insight": "**A collision fallback should add meaning, not a counter.** `<Parent>_<NavProperty>` is unique wherever the parent is and reads as the Mendix convention; a counter depends on import order. Re-import dedup matches by RemoteParentNavigationProperty, so existing suffixed names are left alone (verified: second run 0 created, 1 association). Measured on 11.12.6: Integ.Product_Category, mx check 0 errors."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl/executor", "date": "2026-11-18", "refs": ["ako/mxcli#1258"], "ce": ["CE6630"], "symptom": "`create external entities from` over a contract whose entity set has FilterRestrictions Filterable=false, or NonFilterableProperties naming the key, built an app mx check rejected with one CE6630 \"'Id' is marked Filterable=False in the OData service, but True in the app\" per such entity. The key attribute's own Filterable was already false; every other property was right.", "file": "sdk/domainmodel RemoteKeyPart.NotFilterable; mdl/executor/cmd_contract.go (key parts get !entitySet.AttrFilterable(keyName)); mdl/backend/modelsdk/domainmodel_write.go odataKeyPartToGen (was SetFilterable(true)) and domainmodel.go odataKeyFromGen/storedKeyPartFilterable. Tests: cmd_contract_key_filterable_test.go, odata_key_filterable_test.go; fixture mdl-examples/bug-tests/key-filterable-from-contract.mdl + contracts/key-filterable-metadata.xml", "cause": "A Rest$ODataKeyPart carries its own Filterable beside the attribute's Rest$ODataMappedValue.Filterable, and mxbuild compares both with the service. The writer hard-coded the key part's to true, so the four earlier CE6630 fixes (set-level and property-list shapes of Filterable, Sortable, Top, Skip, Updatable) all set the attribute and none could reach the key part. The reader never read it, so a rewrite of a Studio Pro entity would also have flipped a stored false to true.", "insight": "**Five CE6630 records were the importer getting a capability wrong one term at a time; the efficient way to close the class was one contract per capability shape against the real build, not another report.** 13 shapes, one contract and one project each (whole set Sortable/Filterable=false, NonSortable/NonFilterableProperties naming the label or the key, TopSupported, SkipSupported, CountRestrictions, Insert/Update/DeleteRestrictions, no annotation): 11 clean, 2 CE6630 - both on the KEY and both Filterable. Sortable on the key was already fine, which is what pointed away from the contract parser (AttrFilterable and AttrSortable are the same shape) and at a field only the key has. A model field with a historic default needs a zero value that keeps the default: `NotFilterable` rather than `Filterable`, so every other builder of a key part is unchanged. The reader must not read an absent property as false (storedKeyPartFilterable defaults true), the same trap storedLocalizeDate documents. A hand-written `create external entity` is NOT a CE6630 gap worth a check: it produces CE6010 (no key), CE6612 on every attribute and Countable/Top/Skip mismatches whatever the contract says, so it is not a form that builds on its own. My first matrix also forgot ODataVersion: OData4 (CE6631 on everything) and used the reserved attribute name `Id` in a hand-written entity (CE7247); both were mine. Control: reverting the writer, the importer or the reader each fails its own tests (2, 3 and 1); the fixture gives 2 CE6630 with the pre-fix binary and 0 errors with the fixed one."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl/executor", "date": "2026-11-18", "refs": ["ako/mxcli#1258"], "ce": ["CE1613", "CE0277", "CE0282"], "symptom": "An import mapping's custom object handler `find … by M.MF ( P: member )` that names a parameter the microflow does not declare (or the right name in another case) passed `check` and `exec` and was written; `mx check` reported CE1613 \"The selected parameter 'M.MF.P' no longer exists\". Naming only some of the microflow's parameters was CE0277 \"Parameter mapping of microflow for obtaining objects needs to be refreshed\".", "file": "mdl/executor/validate_mapping_handler_params.go ValidateMappingHandlerParams (MDL104), called from validate_program.go with the lazy project opener. Tests: validate_mapping_handler_params_test.go; fixtures mdl-examples/bug-tests/mapping-handler-unknown-parameter(.fail).mdl, mapping-handler-missing-parameter.fail.mdl", "cause": "buildCustomHandler qualifies each parameter against the microflow (`Microflow + \".\" + Parameter`) and never looks the name up, and nothing counts the parameters. CE0281 itself, the code this was started from, cannot be produced by hand: exec refuses a `by` member that is not in the JSON structure (addCustomHandlerValueElements), so mx check never sees it; check does not mirror that refusal and it was left alone because the mirror would duplicate the builder's JSON-path tracking.", "insight": "**One mapping per project when measuring a build that stages its errors: a shared project made two real errors read as clean.** My first matrix put nine mappings in one project; three were CE1613 (an unresolved reference, stage 0, which hides every other error in every document), so `Index: code` (really CE0282) and one-of-two parameters (really CE0277) both showed '-' and I wrote 'leaving a parameter out is clean' into the rule's header and a test. It surfaced only because the control fixture, applied alone, reported 1 error. Re-measured isolated: names exact and all named clean; unknown, wrong-case or extra name CE1613; one or none of the parameters named CE0277; none named on a zero-parameter microflow clean; a String member into an Integer parameter CE0282 (not covered). The two codes need two fail fixtures for the same reason: in one project CE1613 hides the CE0277. An unknown name is judged alone, as the build does. The same staging that the documented stage table already records was the trap, so run the control and every fail fixture in their own project before trusting a count. Sweep: 1,325 repo .mdl files, 0 MDL104 hits. Control: stubbing the rule fails 12 sub-tests."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl-executor", "date": "2026-11-18", "refs": ["ako/mxcli#2353"], "ce": [], "symptom": "`describe contract operation from openapi` showed `Query: ($limit: Integer)` for an integer query parameter, but after `create or modify consumed rest service … ( OpenAPI: … )` describe showed `Query: ($limit: String)`; path parameters kept Integer in both.", "file": "mdl/executor/cmd_rest_clients.go convertOpenAPIToModel. Test: mdl/executor/openapi_query_type_2353_test.go", "cause": "Rest$QueryParameter has no DataType in the metamodel (Name, TestValue, ParameterUsage only), so the writer cannot store a query type and describe reads it back as String; the preview printed the spec's type from the in-memory model instead.", "insight": "**When the preview and the write disagree, check the metamodel before 'fixing' the write.** The import was right — a consumed REST query parameter is text in Mendix — and the preview was the one promising a type no storage can hold. Measured on 11.12.6: preview and imported describe now both show `$limit: String`, `$petId: Integer`. An MDL `Query: ($x: Integer)` is still stored as String without a word; not addressed here."}
Original file line number Diff line number Diff line change
@@ -0,0 +1 @@
{"area": "mdl/linter", "date": "2026-11-18", "symptom": "A new lint rule that reads the MODEL rather than the catalog reports Mendix's own platform content. MPR016's first run on testdata/pedapp reported \"41 entities overlap each other on the System domain model diagram: ConsumedODataConfiguration, Error, FileDocument, HttpHeader, HttpMessage, HttpRequest and 35 more\" — a model no user can edit (`mxcli layout` refuses System outright). The project's own modules were clean.", "cause": "Catalog-backed rules get the platform filter for free: every ctx.Entities()/Pages()/... query ends in notPlatformModule(alias) (mdl/linter/context.go), which is `COALESCE(Source,'')='' AND COALESCE(Id,'')<>'00000000-0000-0000-0000-000000000001'`. A rule reading the model through the reader (here ListDomainModels, for entity Location, which the catalog does not hold) never passes through that WHERE clause, so it starts out seeing System and every Marketplace module. Fixed by adding LintContext.OwnedModuleNamesByID, the Go counterpart of notPlatformModule, and keying the rule's module resolution on it.", "file": "`mdl/linter/context.go` (OwnedModuleNamesByID, notPlatformModule, systemModuleID); `mdl/linter/rules/mpr016_overlapping_entities.go`; test `mdl/linter/rules/mpr016_overlapping_entities_test.go` (TestOverlappingEntitiesRule_SkipsPlatformModules)", "insight": "The platform filter lives in the SQL, so it protects only rules that ask the catalog. Any rule needing geometry, an expression or anything else the catalog does not carry must re-apply it in Go, and every such rule starts out broken in exactly this way — QUAL002 shipped it once on these same entities, and MPR016 reproduced it three months later. Two cheap discriminators: run the new rule next to a catalog-based rule with a similar threshold (MPR003 caps a module at 15 persistent entities and reported NOTHING on the same 41-entity System module, which localises the difference to the data source in one step), and read the finding's own entity names — FileDocument and HttpRequest are Mendix's, not any app's. Do NOT reach for modules.Source to fix it: Source is empty for System exactly as for the user's own modules, which is the trap the ID test exists for. Rules reading the model are about to become common (an attribute read-but-never-written check, a model-name-in-a-string-literal check and an index check all need the model), so the helper is the fix rather than a filter per rule."}
Loading
Loading