fixup: address PR #258 review feedback

- Narrow v1 ContentsHash fallback to actual schema errors (ako)
  Previously any UPDATE error would trigger the "older schema"
  retry, potentially masking real failures (disk full, invalid
  UnitID, rolled-back tx). Now only retries when the error
  message mentions ContentsHash.
- Drop spurious cloneRoleIDs in drop-then-create path (ako)
  consumeDroppedMicroflow already removed the cache entry, so
  the caller owns the slice — no need to clone again.
- Add mdl-examples/bug-tests/258-drop-create-microflow-unitid-preservation.mdl
  Bug reproducer script as required by CLAUDE.md checklist.

Author: hjothamendix

mdl-examples/bug-tests/258-drop-create-microflow-unitid-preservation.mdl

@@ -0,0 +1,45 @@
+-- ============================================================================
+-- Bug #258: DROP + CREATE OR MODIFY microflow UnitID preservation
+-- ============================================================================
+--
+-- Symptom (before fix):
+--   A DROP MICROFLOW X followed by CREATE OR MODIFY MICROFLOW X in the same
+--   session produced a *different* UnitID for the new row. Studio Pro treats
+--   Unit rows with a different UnitID as unrelated documents and refuses to
+--   open the project:
+--     ".mpr does not look like a Mendix Studio Pro project file"
+--
+-- After fix:
+--   The executor remembers the dropped UnitID (and ContainerID, AllowedRoles)
+--   and reuses them on the subsequent CREATE so Studio Pro sees an in-place
+--   update. Also exercises v1 ContentsHash persistence — MPR v1 projects need
+--   the hash populated on every write or Studio Pro rejects the file as
+--   corrupt.
+--
+-- Usage:
+--   mxcli exec mdl-examples/bug-tests/258-drop-create-microflow-unitid-preservation.mdl -p app.mpr
+--   Open the project in Studio Pro — it must load without corruption errors
+--   and the microflow MF_Replaced must be visible under BugTest258.
+-- ============================================================================
+
+create module BugTest258;
+
+create microflow BugTest258.MF_Original
+    returns string
+(
+    set $return = 'original'
+);
+
+-- Drop + recreate in the same session. Before the fix, this produced a
+-- dangling Unit row with a fresh UUID and the project became unopenable in
+-- Studio Pro. After the fix, the new microflow reuses the original UnitID.
+drop microflow BugTest258.MF_Original;
+
+create or modify microflow BugTest258.MF_Original
+    returns string
+(
+    set $return = 'replaced'
+);
+
+-- DESCRIBE must round-trip cleanly (no placeholder UUIDs leaking).
+describe microflow BugTest258.MF_Original;

mdl/executor/cmd_microflows_create.go

@@ -95,7 +95,9 @@ func execCreateMicroflow(ctx *ExecContext, s *ast.CreateMicroflowStmt) error {
 		if s.Folder == "" && dropped.ContainerID != "" {
 			containerID = dropped.ContainerID
 		}
-		existingAllowedRoles = cloneRoleIDs(dropped.AllowedRoles)
+		// consumeDroppedMicroflow removed the cache entry, so we own this
+		// slice — no need to clone it again.
+		existingAllowedRoles = dropped.AllowedRoles
 		preserveAllowedRoles = true
 	}
 

sdk/mpr/writer_core.go

@@ -144,7 +144,9 @@ func (wt *WriteTransaction) WriteUnit(unitID string, contents []byte) error {
 	_, err := wt.tx.Exec(`
 		UPDATE Unit SET Contents = ?, ContentsHash = ? WHERE UnitID = ?
 	`, contents, contentsHash, unitIDBlob)
-	if err != nil {
+	if err != nil && isContentsHashSchemaError(err) {
+		// Older v1 schemas do not have ContentsHash; retry without it.
+		// Any other error (disk full, invalid UnitID, rolled-back tx) propagates.
 		_, err = wt.tx.Exec(`
 			UPDATE Unit SET Contents = ? WHERE UnitID = ?
 		`, contents, unitIDBlob)

sdk/mpr/writer_units.go

@@ -9,10 +9,24 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
+	"strings"
 
 	"github.com/mendixlabs/mxcli/sdk/domainmodel"
 )
 
+// isContentsHashSchemaError returns true when the error looks like SQLite complaining
+// about the absence of the ContentsHash column (i.e. an old MPR v1 schema from pre-Mx
+// versions that predate ContentsHash). Anything else — a disk-full error, a missing
+// UnitID, a rolled-back transaction — must propagate so writes don't silently succeed
+// without updating Contents.
+func isContentsHashSchemaError(err error) bool {
+	if err == nil {
+		return false
+	}
+	msg := err.Error()
+	return strings.Contains(msg, "ContentsHash")
+}
+
 // updateTransactionID updates the _Transaction table with a new UUID.
 // Studio Pro uses this to detect external changes during F4 sync.
 // Only applies to MPR v2 projects (Mendix >= 10.18).
@@ -154,8 +168,9 @@ func (w *Writer) updateUnit(unitID string, contents []byte) error {
 	_, err := w.reader.db.Exec(`
 		UPDATE Unit SET Contents = ?, ContentsHash = ? WHERE UnitID = ?
 	`, contents, contentsHash, unitIDBlob)
-	if err != nil {
-		// Older v1 schemas do not have ContentsHash.
+	if err != nil && isContentsHashSchemaError(err) {
+		// Older v1 schemas do not have ContentsHash; retry without it.
+		// Any other error (disk full, invalid UnitID, rolled-back tx) propagates.
 		_, err = w.reader.db.Exec(`
 			UPDATE Unit SET Contents = ? WHERE UnitID = ?
 		`, contents, unitIDBlob)