Problem

When creating a cohort database from a prototype on a different DBMS type (e.g., SQL Server → MySQL or PostgreSQL), the operation fails with collation errors.

Current Behavior

The code in CreateNewCohortDatabaseWizard.cs (lines 116-132) attempts to preserve collation from source columns, but:

• DBMS-specific collation names (e.g., Latin1_General_CI_AS on SQL Server vs latin1_swedish_ci on MySQL) are not compatible across systems
• No translation or fallback logic exists for cross-DBMS scenarios
• Results in errors like "collation 'latin1_swedish_ci' not found" when crossing DBMS boundaries

Expected Behavior

When target database DBMS differs from source column DBMS:

• Either set Collation = null to use target database defaults
• Or implement collation name translation between DBMS types

Related Files

• Rdmp.Core/CohortCommitting/CreateNewCohortDatabaseWizard.cs:116-132
• Rdmp.Core/Curation/Data/ColumnInfo.cs (Collation property)

Upstream Reference

Originally reported in HicServices/RDMP#1093 (marked high priority)

Suggested Fix

Add DBMS type check before applying collation:

// Only use collation if source and target are same DBMS type
var sourceDbms = e.ColumnInfo?.TableInfo?.Server?.DatabaseType;
var targetDbms = _targetDatabase.Server.DatabaseType;
Collation = (collations.Length == 1 && sourceDbms == targetDbms) ? collations[0] : null