Repository navigation
Sync ako/mxcli: drop the v1 BSON driver; container RenderMode, Calendar expression properties (CE0642), column filter-slot hint - #1594
Merged
Merged
Conversation
A layout or page container's RenderMode was stored but describe never printed it, so it read as dropped and a describe -> exec round trip did drop it. alter set routed RenderMode only for dynamic texts and refused it on a container, and create wrote any string verbatim into the enum. Fixes mendixlabs#1582 (#1487). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SPETbWEhRQ85EX4gckfCxU
The Marketplace Calendar's required editable and showEventDate expressions (defaultValue="true") failed mx check with CE0642: the mpk-generated template wrote them empty, the def generator gave a top-level expression no mapping, editable: only reached the widget-level editability (which the Calendar does not declare), and describe printed neither. The template now writes an expression's declared default, the generator maps every top-level expression by its own key (v19), editable: routes to the widget's own property when the package declares no Editability, and describe prints stored expressions. Fixes mendixlabs#1578 (#1483). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SPETbWEhRQ85EX4gckfCxU
go.mongodb.org/mongo-driver v1.17.10 is removed. mxcli now runs on mongo-driver/v2 v2.9.2 alone, and `go version -m bin/mxcli` lists only v2. It was the last component left on a customer's blocking dependency scan, flagged for CVE-2026-107325, which has no fix on the v1 line and never will: v1.17.10 is that line's final release and the module is marked deprecated. mxcli was never exposed — the panic is in bsoncore.Array.Validate, which no file imported — but the finding would have recurred on every scan, with no upgrade able to clear it. The migration is not a find-and-replace. The two drivers decode nested documents into different Go types, and the type assertions mxcli reads them back with were written against v1's. Measured on identical bytes: destination v1 nested v2 nested bson.D primitive.D bson.D equivalent bson.M primitive.M bson.D BREAKS map[string]any map[string]any bson.D BREAKS A failed type assertion is not an error, so the value is silently dropped and the next write emits a document with a property missing, which Studio Pro refuses to merge (canon.CompletePropertySets, mendixlabs#1373). The repo's own tests caught every instance: scheduledevents wrote 13 properties where Studio Pro writes 14, TestNoopRerun_ReportsUnchanged tripped the settings guard, and the Atlas layout copies lost their icon and image lookups. Rather than audit the 618 type assertions in the migrated files — where a wrong one compiles clean — this uses the switch v2's decoder already has. New leaf package modelsdk/bsoncompat: UnmarshalDocM DefaultDocumentM() -> nested bson.M, v1's shape UnmarshalDocMap DefaultDocumentMap() -> nested map[string]any DMap -> the bson.D.Map() v2 removed Applied at 73 call sites found by how each Unmarshal target was DECLARED, not by reading the call: 39 bson.M destinations and 34 map[string]any. The two helpers are not interchangeable — bson.M is a named type, so a bson.M value does not satisfy .(map[string]any) — and a bson.D destination needs neither, since a helper there re-introduces the bug inverted. mdl/executor's TestResolveWebServiceName_ReadsMapDecodedDocuments already existed for that exact confusion and caught this migration too, because it pins its own precondition instead of passing vacuously. bsoncompat depends on nothing inside mxcli. It cannot live in modelsdk/codec: codec imports modelsdk/mpr, which imports canon and meta, whose own tests could then not import codec at all ("import cycle not allowed in test"). Guards, both verified by control (each names the offending file or go.mod line): TestNoV1DriverImports walks the tree for the v1 import and TestGoModDoesNotRequireV1Driver reads go.mod, since re-adding either is not a build error and nothing else would notice until the next scan. TestDecodeShapes pins all three nested shapes and fails loudly if a driver release makes a helper unnecessary. Note `go mod tidy` cannot run in this repo at all — mdl/exprcheck/ roundtrip_test.go (//go:build roundtrip) imports the long-deleted mdl/backend/mpr — so the requirement was dropped with `go mod edit -droprequire` and a manual go.sum prune. That dead test file is pre-existing and left untouched here. Verified: go build ./... clean, go vet clean, gofmt clean, 105 test packages pass with 0 failures, `go mod verify` reports all modules verified, and the built binary links only mongo-driver/v2 v2.9.2. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Rer1mzP69XxbXKxQqheipM
A Data grid 2 column has no Filterable property: its filter is the column's filter slot, written as a filter widget in its braces. check already refused the key, but the hint listed the column's properties (with a project) or suggested FilterType (without one). A key that starts with an item's widgets-slot key now names the slot, the body form, and the widget types it routes. Fixes the remaining part of mendixlabs#177 (#1230). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SPETbWEhRQ85EX4gckfCxU
Read the store before the writer (a container's RenderMode was stored; only describe lacked it), and a builtin key can shadow a pluggable widget's own property of the same name (the Calendar's editable). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01SPETbWEhRQ85EX4gckfCxU
Merging origin/main brought in PR #1495 (ALTER NAVIGATION / ALTER MENU), whose 7 new and changed files import the v1 BSON driver. The merge itself was conflict-free, so nothing flagged them — which is exactly what TestNoV1DriverImports is for: it failed on the merged tree and named them. All 7 decode into bson.D destinations, which need no bsoncompat helper: v2 nests bson.D inside bson.D just as v1 nested primitive.D, so their type assertions are already correct. The change is the import path and primitive.X -> bson.X only. Verified on the merged tree: go build ./... clean, go vet clean, gofmt clean, 106 test packages pass with 0 failures, zero v1 imports remain, and the built binary links only mongo-driver/v2 v2.9.2. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Rer1mzP69XxbXKxQqheipM
mdl/exprcheck/roundtrip_test.go could not compile. It imported three things that no longer exist: mdl/backend/mpr (the legacy engine, deleted), executor.GetHierarchyForMining and executor.DescribeMicroflowToString. It never broke a build because it sat behind `//go:build roundtrip` — a tag no Makefile target and no CI job has ever passed, and the only file in the tree that uses it. So it rotted through the legacy-engine deletion and two executor API removals without one red build. The only command that still saw it was `go mod tidy`, which considers every build tag and failed with "no matching versions for query latest" on the deleted package — a message that reads as a dependency problem rather than a dead test. That is why the v1 BSON driver had to be dropped from go.mod with `go mod edit -droprequire` and a hand-pruned go.sum earlier on this branch. Ported to mdl/executor/exprcheck_roundtrip_test.go under the `integration` tag rather than deleted: the assertion is one the unit tests cannot make. Describing every microflow and re-parsing each expression with the robust parser catches grammar gaps — a missing parsePrimary branch, a missing funcTable entry, a missing slot constraint — where the unit tests only ever feed the parser expressions someone thought to write down. It lives in package executor because that is where the real-project harness already is (setupTestEnv, describeMDL) and because `make test-integration-executor` actually runs it. The import cycle the original file's header warned about does not arise in this direction: mdl/exprcheck does not import mdl/executor. The removed GetHierarchyForMining is replaced by building the containerID -> module name map from backend.ListModules(). Two vacuity guards added, because the fixture project lives in a separate repo (mx-test-projects/) and TestMain skips the whole suite when it is absent: the test now fails if no microflow could be described or if no expression slot was found, rather than reporting success against an empty walk. With the dead import gone, `go mod tidy` runs clean and pruned 35 more stale go.sum lines than the hand-edit had, so the v1 driver's removal is now the tool's work rather than mine. Verified: go build ./... clean, go vet clean (also with -tags integration), gofmt clean, 106 test packages pass with 0 failures, `go mod tidy` exits 0, `go mod verify` reports all modules verified. The ported test compiles under the integration tag but SKIPS in this container, which has no mx-test-projects checkout — CI's `integration (executor)` job is what exercises it. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Rer1mzP69XxbXKxQqheipM
CI's integration (executor) job failed the round-trip test ported in the previous commit, and failed it on the vacuity guard rather than on a hint: described 1 microflow(s) but found no expression slots; either the test project lost its expressions or the walk below no longer matches the AST The guard was right and the test was wrong. A microflow's ContainerID is its FOLDER when it sits in one, not its module, so the flat map I built from ListModules resolved almost nothing — and an unresolved module name meant the microflow was skipped before it was ever described. That is precisely the job the removed executor.GetHierarchyForMining did, and replacing it with a flat lookup is what broke the walk. NewContainerHierarchyFromBackend is the live replacement: FindModuleID walks up through folders to the owning module. It is exported and this test is in package executor, so it needs no new seam. Measured locally against testdata/expr-checker/minimal.mpr, which this container does have (the integration fixture, mx-test-projects/, it does not — hence the first push going out unvalidated): resolved by flat ListModules map : 1/16 microflows resolved by hierarchy FindModuleID: 16/16 microflows and the full describe -> reparse -> walk chain then yields 16 microflows described and 11 SourceExpr slots, where it previously found 0. The other 36 expression nodes are LiteralExpr and VariableExpr, which the walk deliberately ignores: they are already structured, so there is no raw source text for the robust parser to re-check. So both halves of the guard's complaint are closed, and the thresholds it asserts (at least one microflow described, at least one expression slot) are met with margin rather than exactly. Verified: go build ./... clean, go vet clean (also with -tags integration), gofmt clean, 106 test packages pass with 0 failures. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Rer1mzP69XxbXKxQqheipM
…55o66g fix: silent-property-drop cluster — container RenderMode, Calendar expressions, column Filterable
refactor(deps): move off the deprecated v1 BSON driver entirely Removes go.mongodb.org/mongo-driver v1.17.10; mxcli now runs on mongo-driver/v2 v2.9.2 alone. It was the last component on a customer's blocking dependency scan, flagged for CVE-2026-107325, which has no fix on the v1 line and never will — v1.17.10 is that line's final release and the module is deprecated. The two drivers decode nested documents into different Go types, so a bson.M or map[string]any destination silently stopped matching its type assertions and dropped properties from the next write. New leaf package modelsdk/bsoncompat (UnmarshalDocM, UnmarshalDocMap, DMap) restores v1's shapes at 73 call sites; a bson.D destination needs no helper. Three guards keep v1 from returning, since re-adding it is not a build error. Also ports mdl/exprcheck/roundtrip_test.go, dead behind a build tag no CI job ran, to mdl/executor under the integration tag — which unblocked `go mod tidy` for the whole repo. Validated beyond CI: tested against the Formula1 solution (two Mendix 11.14.0 apps, 1,060 units, 41 MDL scripts) with base as a control, and on five further test projects. No regression; writes byte-identical to main wherever bytes are deterministic, and every unit keeps the property names Studio Pro writes.
main is red. The merge of #1497, which removed go.mongodb.org/mongo-driver v1 from go.mod, landed just after #1498 added mdl/backend/pagemutator/container_rendermode_test.go — a file written against v1. The package now fails to compile at all: container_rendermode_test.go:8:2: no required module provides package go.mongodb.org/mongo-driver/bson FAIL github.com/mendixlabs/mxcli/mdl/backend/pagemutator [setup failed] and TestNoV1DriverImports names the file, which is what it is for. Neither PR is at fault. #1498 was authored and tested against a main that still required v1, the two touched different files so nothing conflicted, and each was green on its own base. The combination is what breaks. The file uses bson.D only, so this is the import path and nothing else: v2 nests bson.D inside bson.D exactly as v1 nested primitive.D, so its assertions are already correct and it needs no modelsdk/bsoncompat helper. Checked the whole tree rather than just the file CI named, since `go test` stops at the first [setup failed] package and a second offender would only surface on the next run: this is the only one. Verified: go build ./... clean, go vet clean, gofmt clean, 106 test packages pass with 0 failures, the two packages that failed in CI (mdl/backend/pagemutator and modelsdk/bsoncompat) both pass, and `go mod tidy` exits 0 leaving go.mod and go.sum unchanged — so v1 stays out. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Rer1mzP69XxbXKxQqheipM
fix(test): point container_rendermode_test at the v2 BSON driver Unbreaks main. #1497 removed go.mongodb.org/mongo-driver v1 from go.mod just after #1498 added a test file written against it, so mdl/backend/pagemutator stopped compiling and TestNoV1DriverImports named the file. Neither PR is at fault: each was green on its own base and the two touched different files, so nothing conflicted — only the combination breaks. The file uses bson.D only, so this is the import path alone; v2 nests bson.D inside bson.D exactly as v1 nested primitive.D, needing no modelsdk/bsoncompat helper. Whole tree checked, not just the file CI named: this is the only one.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Syncs
ako/mxcli:maininto upstream: 9 commits from fork PRs ako#1498 (silent-property-drop fixes), ako#1497 (dropping the v1 BSON driver) and ako#1500 (a test follow-up).Fixes
create layout: a container'sRenderModeis silently dropped; check and exec pass #1582).bson dumpshowsRenderMode: "Main"stored. Butdescribenever printed it, so it looked dropped, and a describe → exec round trip really did drop it.describenow printsRenderMode: Main, leaving out the defaultDiv.alter page|layout … set (RenderMode: …)on a container now writes the value. Before, it was refused as "not a property of this built-in widget".createused to write any string verbatim into the enum ('main','Banner'). Values are now matched case-insensitively against the generated enum, and an unknown member is refused.editableexpression cannot be written;mx checkfails CE0642 #1578). The Marketplace Calendar's requirededitableandshowEventDateused to failmx checkwith CE0642. Four gaps are closed:.mpknow writes an expression's declareddefaultValueinstead of an empty expression.Editabilitysystem property,editable:goes to the widget's own property. check (MDL-WIDGET41) and the build both follow that.describeprints stored expressions.Filterable:hint names the filter slot (feat: SupportFilterableproperty on DataGrid2 COLUMN in MDL #177, the remaining part). check already refused the key, but its hint listed the column's properties, or suggestedFilterType. It now explains thatfilteris a slot and shows the body form:column (…) { textfilter w1 (…) }.Dependencies: the deprecated v1 BSON driver is gone (ako#1497)
go.mongodb.org/mongo-driverv1.17.10 is removed. mxcli now runs onmongo-driver/v2v2.9.2 alone. The v1 module was flagged for CVE-2026-107325, which has no fix on the v1 line. mxcli was never exposed to it, but no upgrade could clear the scan finding.bson.Dwhere v1 gaveprimitive.Mormap[string]any. So a.(bson.M)or.(map[string]any)type assertion silently fails, and the next write drops a property.modelsdk/bsoncompat(UnmarshalDocM,UnmarshalDocMap,DMap) uses v2's own decoder options to restore v1's shapes.bson.Ddestinations need no change.TestNoV1DriverImportsandTestGoModDoesNotRequireV1Drivercatch v1 coming back. The first already caught 7 files from ALTER NAVIGATION/MENU, which were migrated in a follow-up commit.TestDecodeShapespins the nested shapes, so it fails loudly if a driver release makes a helper unnecessary.go mod tidyruns again.mdl/exprcheck/roundtrip_test.gosat behind aroundtriptag that nothing ever passed, and it imported the deleted legacy engine. It is ported tomdl/executor/exprcheck_roundtrip_test.gounderintegration, with a vacuity guard. A follow-up commit makes it resolve a microflow's module through folders.checkpasses builds that fail with CE1406 and CE5400 #1498's container RenderMode test now uses the v2 driver (fix(test): point container_rendermode_test at the v2 BSON driver ako/mxcli#1500). The two PRs landed side by side, and the test still imported v1.Validation
mxcli newapp with Calendar v2.6.1:maingives CE0642 twice. This branch gives 0 errors with the properties unset, set to'true', and set to false.Unchanged page.mdl-examples/bug-tests/. The Calendar one needs the package.make build,test,lint,check-mdl,check-findingsandcheck-wiki-pagespass.go build,go vet(also under-tags integration),go mod tidyandgo mod verifypass.go version -m bin/mxclilists onlymongo-driver/v2.Docs
docs-wiki/bug-patterns/silent-property-drop.mdgets two insights: