feat: add CREATE/DROP NANOFLOW support by retran · Pull Request #7 · retran/mxcli

retran · 2026-04-23T15:42:41Z

Why

mxcli supports CREATE/DROP for microflows but not nanoflows, leaving a gap in the MDL surface. Nanoflows are a core Mendix document type used heavily in client-side logic. Without this, users must fall back to Studio Pro for any nanoflow scaffolding — breaking the CLI-first workflow. Additionally, nanoflows have a restricted action palette and return type constraints that differ from microflows; these must be validated at CREATE time to prevent invalid models.

Summary

Add createNanoflowStatement grammar rule in MDLParser.g4, reusing microflowParameterList, microflowReturnType, microflowOptions, and microflowBody rules
Add CreateNanoflowStmt and DropNanoflowStmt AST types
Add ExitCreateNanoflowStatement visitor and NANOFLOW branch in ExitDropStatement
Add execCreateNanoflow and execDropNanoflow executor handlers mirroring microflow pattern (without AllowedModuleRoles, AllowConcurrentExecution, ReturnVariableName — nanoflows don't have these fields)
Add nanoflow cache fields (createdNanoflows, droppedNanoflows) and tracking helpers for DROP+CREATE rewrite safety
Add nanoflow_validation.go with nanoflow-specific validation:
- Reject 21 microflow-only action types (Java actions, REST calls, workflow actions, import/export, external actions, show home page, JSON transform, ErrorEvent)
- Reject Binary return type (nanoflows cannot return Binary)
- Recursive validation through compound statements (if/loop/while) and error handling bodies
Register nanoflow handlers in registerMicroflowHandlers
Update allKnownStatements test snapshot

Stacked on

refactor: add type-safe BSON helpers in writer_security #6 (pr3-type-assertion-hardening)

Test

make build && make test && make lint-go — all pass

Copilot

Pull request overview

Adds end-to-end MDL support for creating and dropping nanoflows, mirroring the existing microflow pipeline (grammar → AST → visitor → executor registry/handlers), so MDL scripts can manage nanoflow documents in Mendix projects.

Changes:

Extend the MDL grammar + generated parser listeners to parse CREATE NANOFLOW and DROP NANOFLOW.
Introduce CreateNanoflowStmt / DropNanoflowStmt AST nodes and corresponding visitor logic.
Add executor handlers for create/drop nanoflows, plus executor cache fields for created/dropped nanoflow tracking and registry wiring.

Reviewed changes

Copilot reviewed 13 out of 15 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
mdl/grammar/MDLParser.g4	Adds `createNanoflowStatement` and wires it into `createStatement`.
mdl/grammar/parser/mdlparser_listener.go	Regenerated listener interfaces to include nanoflow enter/exit hooks.
mdl/grammar/parser/mdlparser_base_listener.go	Regenerated base listener stubs for the new nanoflow production.
mdl/grammar/parser/mdl_lexer.go	Regenerated lexer header (ANTLR version bump).
mdl/ast/ast_microflow.go	Adds AST statement types for `CREATE/DROP NANOFLOW`.
mdl/visitor/visitor_microflow.go	Builds `CreateNanoflowStmt` from parse tree (mirrors microflow builder).
mdl/visitor/visitor_entity.go	Adds `NANOFLOW` branch to `ExitDropStatement`.
mdl/executor/register_stubs.go	Registers create/drop nanoflow handlers (currently under microflow handler registration).
mdl/executor/cmd_nanoflows_create.go	Implements `execCreateNanoflow` (mirrors microflow create flow builder + validations).
mdl/executor/cmd_nanoflows_drop.go	Implements `execDropNanoflow` (mirrors microflow drop behavior + dropped-ID tracking).
mdl/executor/executor.go	Extends executor cache structs and adds dropped-nanoflow tracking helpers.
mdl/executor/exec_context.go	Adds nanoflow “created during session” tracking helper.
mdl/executor/registry_test.go	Updates the known-statement snapshot list to include nanoflow statements.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-23T15:49:40Z

+createNanoflowStatement
+    : NANOFLOW qualifiedName
+      LPAREN microflowParameterList? RPAREN
+      microflowReturnType?
+      microflowOptions?
+      BEGIN microflowBody END SEMICOLON? SLASH?
+    ;


createNanoflowStatement is introduced without the explanatory doc block/examples that createMicroflowStatement has immediately above. If this grammar file is intended to be user-facing documentation as well (as it is for other statements), consider adding a short comment + example(s) for nanoflows to keep the CREATE statement docs consistent.

Copilot · 2026-04-23T15:49:41Z

+	// Build entity resolver function for parameter/return types
+	entityResolver := func(qn ast.QualifiedName) model.ID {
+		dms, err := ctx.Backend.ListDomainModels()
+		if err != nil {
+			return ""
+		}
+		modules, _ := ctx.Backend.ListModules()
+		moduleNames := make(map[model.ID]string)
+		for _, m := range modules {
+			moduleNames[m.ID] = m.Name
+		}


entityResolver silently swallows backend errors (ListDomainModels returns "" on error; ListModules error is ignored). This can surface as a misleading "entity not found" validation error when the real issue is a backend read failure. Prefer returning/propagating the backend error so users get an actionable failure reason.

Suggested change

// Build entity resolver function for parameter/return types

entityResolver := func(qn ast.QualifiedName) model.ID {

dms, err := ctx.Backend.ListDomainModels()

if err != nil {

return ""

}

modules, _ := ctx.Backend.ListModules()

moduleNames := make(map[model.ID]string)

for _, m := range modules {

moduleNames[m.ID] = m.Name

}

// Load metadata needed by the entity resolver up front so backend read

// failures are returned as actionable errors instead of being treated as

// "entity not found".

dms, err := ctx.Backend.ListDomainModels()

if err != nil {

return mdlerrors.NewBackend("list domain models", err)

}

modules, err := ctx.Backend.ListModules()

if err != nil {

return mdlerrors.NewBackend("list modules", err)

}

moduleNames := make(map[model.ID]string)

for _, m := range modules {

moduleNames[m.ID] = m.Name

}

// Build entity resolver function for parameter/return types

entityResolver := func(qn ast.QualifiedName) model.ID {

Copilot · 2026-04-23T15:49:41Z

+// execCreateNanoflow handles CREATE NANOFLOW statements.
+func execCreateNanoflow(ctx *ExecContext, s *ast.CreateNanoflowStmt) error {
+	if !ctx.ConnectedForWrite() {
+		return mdlerrors.NewNotConnectedWrite()
+	}


execCreateNanoflow is very close to execCreateMicroflow (module/folder resolution, validation, flow graph build, create-or-replace semantics). Consider extracting shared helpers to reduce duplication and prevent microflow/nanoflow behavior drifting apart as fixes/features land over time.

Copilot · 2026-04-23T15:49:42Z

+// execCreateNanoflow handles CREATE NANOFLOW statements.
+func execCreateNanoflow(ctx *ExecContext, s *ast.CreateNanoflowStmt) error {
+	if !ctx.ConnectedForWrite() {
+		return mdlerrors.NewNotConnectedWrite()
+	}


There are extensive integration roundtrip tests for CREATE/DROP MICROFLOW, but this PR adds CREATE/DROP NANOFLOW without similar coverage. Adding at least one integration test that creates a nanoflow (and CREATE OR REPLACE) and then DESCRIBEs/drops it would help prevent regressions in the new executor handlers and dropped-ID reuse logic.

Copilot · 2026-04-23T15:49:42Z

+// trackCreatedNanoflow registers a nanoflow created during this session.
+func (ctx *ExecContext) trackCreatedNanoflow(moduleName, nfName string, id, containerID model.ID, returnEntityName string) {
+	ctx.ensureCache()
+	if ctx.Cache.createdNanoflows == nil {
+		ctx.Cache.createdNanoflows = make(map[string]*createdNanoflowInfo)
+	}
+	qualifiedName := moduleName + "." + nfName
+	ctx.Cache.createdNanoflows[qualifiedName] = &createdNanoflowInfo{
+		ID:               id,
+		Name:             nfName,
+		ModuleName:       moduleName,
+		ContainerID:      containerID,
+		ReturnEntityName: returnEntityName,
+	}


trackCreatedNanoflow populates ctx.Cache.createdNanoflows, but there is no corresponding read path (no getCreatedNanoflow/resolver usage in the repo), so this cache currently adds state without affecting behavior. Either wire this cache into the relevant nanoflow resolvers (similar to createdMicroflows) or remove it to avoid dead code and future divergence.

Suggested change

// trackCreatedNanoflow registers a nanoflow created during this session.

func (ctx *ExecContext) trackCreatedNanoflow(moduleName, nfName string, id, containerID model.ID, returnEntityName string) {

ctx.ensureCache()

if ctx.Cache.createdNanoflows == nil {

ctx.Cache.createdNanoflows = make(map[string]*createdNanoflowInfo)

}

qualifiedName := moduleName + "." + nfName

ctx.Cache.createdNanoflows[qualifiedName] = &createdNanoflowInfo{

ID: id,

Name: nfName,

ModuleName: moduleName,

ContainerID: containerID,

ReturnEntityName: returnEntityName,

}

// trackCreatedNanoflow is kept for API compatibility, but nanoflow creations

// are not cached because there is currently no corresponding read path for

// this session state.

func (ctx *ExecContext) trackCreatedNanoflow(moduleName, nfName string, id, containerID model.ID, returnEntityName string) {

Copilot · 2026-04-23T15:49:42Z

 }
+
+// rememberDroppedNanoflow records the UnitID and ContainerID of a nanoflow
+// that was just deleted so a subsequent CREATE OR REPLACE/MODIFY can reuse them.


The comment for rememberDroppedNanoflow says the nanoflow "was just deleted", but this function is called before deletion (to capture IDs). Consider aligning the wording with rememberDroppedMicroflow ("about to be deleted") to avoid misleading future maintainers.

Suggested change

// that was just deleted so a subsequent CREATE OR REPLACE/MODIFY can reuse them.

// that is about to be deleted so a subsequent CREATE OR REPLACE/MODIFY can reuse them.

Copilot · 2026-04-23T15:49:42Z

 func registerMicroflowHandlers(r *Registry) {
 	r.Register(&ast.CreateMicroflowStmt{}, func(ctx *ExecContext, stmt ast.Statement) error {
 		return execCreateMicroflow(ctx, stmt.(*ast.CreateMicroflowStmt))
 	})
 	r.Register(&ast.DropMicroflowStmt{}, func(ctx *ExecContext, stmt ast.Statement) error {
 		return execDropMicroflow(ctx, stmt.(*ast.DropMicroflowStmt))
 	})
+	r.Register(&ast.CreateNanoflowStmt{}, func(ctx *ExecContext, stmt ast.Statement) error {
+		return execCreateNanoflow(ctx, stmt.(*ast.CreateNanoflowStmt))
+	})
+	r.Register(&ast.DropNanoflowStmt{}, func(ctx *ExecContext, stmt ast.Statement) error {
+		return execDropNanoflow(ctx, stmt.(*ast.DropNanoflowStmt))
+	})
 }


registerMicroflowHandlers now registers nanoflow handlers as well. To keep the registry organization self-descriptive, consider either renaming this function (e.g., microflow/nanoflow) or splitting nanoflow registration into its own helper alongside the other domain-specific registration functions.

Copilot

Pull request overview

Copilot reviewed 13 out of 15 changed files in this pull request and generated 4 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-23T19:49:10Z

+// nanoflow bodies. These correspond to microflow-only actions in the Mendix
+// runtime: Java actions, REST/web service calls, workflow actions, import/export,
+// external object operations, download, push-to-client, show home page, and
+// JSON transformation.


The comment above nanoflowDisallowedActions lists several disallowed categories (e.g., “download”, “push-to-client”, “external object operations”) that are not represented in the actual nanoflowDisallowedActions map. This makes the documentation misleading; either expand the map to include the missing AST statement types (if they exist in this codebase) or adjust the comment to only describe what is actually enforced.

Suggested change

// nanoflow bodies. These correspond to microflow-only actions in the Mendix

// runtime: Java actions, REST/web service calls, workflow actions, import/export,

// external object operations, download, push-to-client, show home page, and

// JSON transformation.

// nanoflow bodies. These correspond to disallowed actions enforced here:

// error events, Java actions, database queries, external action calls,

// REST/web service calls, workflow actions, import/export mappings,

// show home page, and JSON transformation.

Copilot · 2026-04-23T19:49:11Z

+// validateNanoflowReturnType checks that the return type is allowed for nanoflows.
+// Binary and Float return types are not supported.
+func validateNanoflowReturnType(retType *ast.MicroflowReturnType) string {
+	if retType == nil {
+		return ""
+	}
+	switch retType.Type.Kind {
+	case ast.TypeBinary:
+		return "Binary return type is not allowed in nanoflows"
+	}


validateNanoflowReturnType’s doc comment says “Binary and Float return types are not supported”, but the implementation only rejects ast.TypeBinary (and there is no TypeFloat in ast.DataTypeKind). Please either update the comment to match the actual rule, or add the missing return-type validation if another kind is intended to be disallowed.

Copilot · 2026-04-23T19:49:11Z

+func validateNanoflowStatements(stmts []ast.MicroflowStatement, errors *[]string) {
+	for _, stmt := range stmts {
+		typeName := fmt.Sprintf("%T", stmt)
+		if reason, disallowed := nanoflowDisallowedActions[typeName]; disallowed {
+			*errors = append(*errors, reason)
+			continue


Using fmt.Sprintf("%T", stmt) + a map[string]string for disallowed-action matching is brittle (renames/refactors can silently break validation) and harder to make exhaustive. Consider switching to a type switch (grouping disallowed concrete types) or using reflect.Type keys so the compiler helps keep this list accurate.

Copilot · 2026-04-23T19:49:11Z

+	// Validate nanoflow-specific constraints before building the flow graph
+	qualName := s.Name.Module + "." + s.Name.Name
+	if errMsg := validateNanoflow(qualName, s.Body, s.ReturnType); errMsg != "" {
+		return fmt.Errorf("%s", errMsg)
+	}


There’s no executor test coverage for the new CREATE NANOFLOW handler and its nanoflow-specific validation (success paths + validation failures). Please add unit tests similar to the existing microflow create/drop mock-backend tests to prevent regressions (including CREATE OR REPLACE/MODIFY ID reuse after DROP).

Copilot AI review requested due to automatic review settings April 23, 2026 15:42

Copilot started reviewing on behalf of retran April 23, 2026 15:43 View session

retran mentioned this pull request Apr 23, 2026

refactor: introduce LintReader interface, decouple linter from sdk/mpr #5

Open

Copilot AI reviewed Apr 23, 2026

View reviewed changes

retran force-pushed the pr4-nanoflows-create-drop branch from db1a8b7 to 2356562 Compare April 23, 2026 15:51

retran force-pushed the pr3-type-assertion-hardening branch from 231606c to fd6530f Compare April 23, 2026 19:42

retran force-pushed the pr4-nanoflows-create-drop branch from 2356562 to 4588cda Compare April 23, 2026 19:42

retran requested a review from Copilot April 23, 2026 19:42

Copilot started reviewing on behalf of retran April 23, 2026 19:43 View session

Copilot AI reviewed Apr 23, 2026

View reviewed changes

retran force-pushed the pr3-type-assertion-hardening branch from fd6530f to a1e8dad Compare April 23, 2026 20:11

retran force-pushed the pr4-nanoflows-create-drop branch from 4588cda to bc2a53a Compare April 23, 2026 20:11

feat: add CREATE/DROP NANOFLOW support — grammar, AST, visitor, executor

dabcdfb

retran force-pushed the pr3-type-assertion-hardening branch from a1e8dad to fd9950e Compare April 23, 2026 20:28

retran force-pushed the pr4-nanoflows-create-drop branch from bc2a53a to dabcdfb Compare April 23, 2026 20:28

	// that was just deleted so a subsequent CREATE OR REPLACE/MODIFY can reuse them.
	// that is about to be deleted so a subsequent CREATE OR REPLACE/MODIFY can reuse them.

Conversation

retran commented Apr 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Why

Summary

Stacked on

Test

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 23, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

retran commented Apr 23, 2026 •

edited

Loading