Avoid InvalidCastException with Oracle number high precision values

[kriewall]

Describe the issue

I'm getting an InvalidCastException when rehydrating numbers with very small / high precision values in an Oracle dB. The limitation appears to actually be due to the fact that Oracle can represent numbers with higher precision than C# (described here); however my main question here is how to prevent NHibernate from inserting values into the dB that have higher precision than C# can handle. If you have recommendations how to work around the intake issue, that would be much appreciated as well.

Expected behavior

• By default, C# precision limits should be respected when inserting doubles into the dB.
• Ideally intake of high precision values would be handled gracefully; however, I recognize you may not have control over this since the issue occurs in the Oracle reader.

As a workaround to prevent overly precise data from getting inserted into the dB, do I need to specify precision and scale everywhere I map a double? Is there a generic way to do this?

Environment

• Windows 10 and Windows 2016 server (observed on both)
• NHibernate
  • Problematic data inserted into production dB under NHibernate 5.2.7
  • InvalidCastException occurs on both 5.2.7 and 5.3.5.
• Oracle dB: 11.2.0.2
• Oracle.ManagedDataAccess: 18.3, but have also tried with 19.10

Additional information

Stack trace:

 at Oracle.ManagedDataAccess.Client.OracleDataReader.GetDecimal(Int32 i)
       at Oracle.ManagedDataAccess.Client.OracleDataReader.GetValue(Int32 i)
       at Oracle.ManagedDataAccess.Client.OracleDataReader.get_Item(Int32 i)
       at NHibernate.Type.NullableType.NullSafeGet(DbDataReader rs, String name, ISessionImplementor session)
       at NHibernate.Type.NullableType.NullSafeGet(DbDataReader rs, String[] names, ISessionImplementor session, Object owner)
       at NHibernate.Type.AbstractType.Hydrate(DbDataReader rs, Object owner)
       at NHibernate.Persister.Entity.AbstractEntityPersister.Hydrate(DbDataReader rs, Object id, Object obj, String[][] suffixedPropertyColumns, ISet`1 fetchedLazyProperties, Boolean allProperties, Int32[] indexes, ISessionImplementor session)
       at NHibernate.Persister.Entity.LoadableExtensions.Hydrate(ILoadable loadable, DbDataReader rs, ISessionImplementor session)
       at NHibernate.Loader.Loader.LoadFromResultSet(DbDataReader rs, Int32 i, ILoadable persister, EntityKey key, LockMode lockMode, ILoadable rootPersister, ISessionImplementor session)
       at NHibernate.Loader.Loader.InstanceNotYetLoaded(DbDataReader dr, EntityKey optionalObjectKey, Object optionalObject, IList hydratedObjects, ISessionImplementor session)
       at NHibernate.Loader.Loader.GetRow(DbDataReader rs, ILoadable[] persisters, EntityKey[] keys, LockMode[] lockModes, Boolean mustLoadMissingEntity, Action`2 cacheBatchingHandler)
       at NHibernate.Loader.Loader.GetRowFromResultSet(DbDataReader resultSet, QueryParameters queryParameters, LockMode[] lockModeArray, Boolean returnProxies, IResultTransformer forcedResultTransformer, QueryCacheResultBuilder queryCacheResultBuilder, Action`2 cacheBatchingHandler)
       at NHibernate.Loader.Loader.DoQuery(ISessionImplementor session, QueryCacheResultBuilder queryCacheResultBuilder)
       at NHibernate.Loader.Loader.DoQueryAndInitializeNonLazyCollections(ISessionImplementor session, QueryCacheResultBuilder queryCacheResultBuilder)
       at NHibernate.Loader.Loader.LoadCollectionBatch(ISessionImplementor session, IType type)
at NHibernate.Loader.Loader.LoadCollectionBatch(ISessionImplementor session, Object[] ids, IType type)
       at NHibernate.Loader.Collection.BatchingCollectionInitializer.Initialize(Object id, ISessionImplementor session)
       at NHibernate.Persister.Collection.AbstractCollectionPersister.Initialize(Object key, ISessionImplementor session)
       at NHibernate.Event.Default.DefaultInitializeCollectionEventListener.OnInitializeCollection(InitializeCollectionEvent event)
       at NHibernate.Impl.SessionImpl.InitializeCollection(IPersistentCollection collection, Boolean writing)
       at NHibernate.Collection.AbstractPersistentCollection.Initialize(Boolean writing)
       at NHibernate.Collection.AbstractPersistentCollection.Read()
       at NHibernate.Collection.Generic.PersistentGenericBag`1.GetEnumerator()
       at System.Linq.Enumerable.WhereSelectEnumerableIterator`2.MoveNext()
       at System.Linq.Enumerable.WhereSelectEnumerableIterator`2.MoveNext()
       at MathNet.Numerics.Statistics.DescriptiveStatistics.ComputeDecimal(IEnumerable`1 data)
       at MathNet.Numerics.Statistics.DescriptiveStatistics..ctor(IEnumerable`1 data, Boolean increasedAccuracy)
       at Locnes.Core.Domain.State.Smoothing.LotHistoryData.GetFinalCheckStdev() in C:\src\locnes\Locnes.Core\Domain\State\Smoothing\LotHistoryData.cs:line 98
       at Locnes.Core.Domain.State.Smoothing.VdewmaStateEngine.<>c__DisplayClass35_0.<AdvanceState>b__0(LotHistoryData d) in C:\src\locnes\Locnes.Core\Domain\State\Smoothing\VdewmaStateEngine.cs:line 78
       at System.Linq.Enumerable.WhereSelectEnumerableIterator`2.MoveNext()
       at System.Collections.Generic.List`1..ctor(IEnumerable`1 collection)
       at System.Linq.Enumerable.ToList[TSource](IEnumerable`1 source)
       at Locnes.Core.Domain.State.Smoothing.VdewmaStateEngine.AdvanceState(EngineData history, IDataStrategyService service) in C:\src\locnes\Locnes.Core\Domain\State\Smoothing\VdewmaStateEngine.cs:line 78
       at Locnes.Core.Domain.State.Smoothing.TermEngine.DoGetAverage() in C:\src\locnes\Locnes.Core\Domain\State\Smoothing\TermEngine.cs:line 132
       at Locnes.Core.Domain.State.Smoothing.TermEngine.<>c__DisplayClass35_0.<GetAverage>b__0() in C:\src\locnes\Locnes.Core\Domain\State\Smoothing\TermEngine.cs:line 118
       at CSharpFunctionalExtensions.Result.Try[T](Func`1 func, Func`2 errorHandler)

SessionFactory config

                var dbConfig = OracleDataClientConfiguration.Oracle10
                    .ConnectionString(c => c.Is(connectionString))
                    .Driver<NHibernate.Driver.OracleManagedDataClientDriver>();
                //.ShowSql();
                dbConfig = ApplicationGlobals.DefaultSchema.Equals(ApplicationGlobals.WindowsUser) ? dbConfig : dbConfig.DefaultSchema(ApplicationGlobals.DefaultSchema);

                FluentConfiguration configuration = Fluently.Configure()
                    .Database(dbConfig)
                    .Mappings(m =>
                    {
                        m.FluentMappings.AddFromAssembly(targetAssembly).AddConventions() /*.ExportTo(@"c:\temp\mappings")*/;
                        //m.FluentMappings.AddFromAssembly(Assembly.GetExecutingAssembly()).AddConventions()/*.ExportTo(@"c:\temp\mappings")*/;
                        //m.HbmMappings.AddFromAssembly(Assembly.GetExecutingAssembly());
                    })
                    .ExposeConfiguration(x =>
                    {
                        x.EventListeners.PostCommitUpdateEventListeners = new IPostUpdateEventListener[] { new EventListener() };
                        x.EventListeners.PostCommitInsertEventListeners = new IPostInsertEventListener[] { new EventListener() };
                        x.EventListeners.PostCommitDeleteEventListeners = new IPostDeleteEventListener[] { new EventListener() };
                        x.EventListeners.PostCollectionUpdateEventListeners = new IPostCollectionUpdateEventListener[] { new EventListener() };
                        x.DataBaseIntegration(dbi =>
                        {
                            dbi.Batcher<OracleDataClientBatchingBatcherFactory>();
                            dbi.Dialect<Oracle10gDialect>();
                            dbi.BatchSize = 100;
                        });
                        x.SetProperty(Environment.DefaultBatchFetchSize, "50");
                        x.Cache(cache => cache.Provider<HashtableCacheProvider>());
                        x.SetInterceptor(new ReferenceDataInterceptor());
                        x.SetInterceptor(new SqlStatementInterceptor());
                    });

I have a separate query routine that uses the Oracle.ManagedDataAccess dll that was also bombing out for an InvalidCastException when used against similar data. I verified using the query that the issue resolved when I changed the value of the problematic number at the last decimal place). I'm still trying to figure out the exact source of the problem under NHibernate, but the signature is basically the same InvalidCastException at GetDecimal().

Example value that causes cast exception: 0.00000000000000422030887989616

Let me know if I can provide any additional information. Appreciate any help you can provide; I'm looking at a pretty ugly issue if I don't get this resolved soon.

[hazzik]

Looks like the issue with the provider.

[kriewall]

Meaning there's nothing I can do from an application perspective that can prevent problematic values from getting inserted into the dB?

[kriewall]

@hazzik I discovered that using GetDouble() allows the value to be returned successfully, as opposed to the GetDecimal() method which gets called from within GetValue(). The property which is having a problem is a nullable double. Is there something about my use of a nullable double that is causing NHibernate to call GetValue() instead of GetDouble(), or does NHibernate use GetValue() across the board?

    <property access="field.camelcase-underscore" name="FinalCheckMeasurement" type="System.Nullable`1[[System.Double, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089]], mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089">
      <column name="FinalCheckMeasurement" not-null="false" />
    </property>

Applying precision and scale doesn't seem to make any difference in how the number is recorded to the dB.

[maca88]

Meaning there's nothing I can do from an application perspective that can prevent problematic values from getting inserted into the dB?

As you are using Oracle 11, you can enable binary floating point types, by setting configuration setting oracle.use_binary_floating_point_types to true. Note that this setting was added in 5.3 by #2349, so it won't work with NHibernate 5.2.7.

[kriewall]

@maca88 Thank you for the suggestion! I tried it out but now I'm seeing InvalidCastExceptions at the same query but across the board for all numeric (double) values. The registration that happens in Oracle10gDialect.RegisterFloatingPointTypeMappings() is applied at data intake, correct? Or is this something I would have had to have enabled before any data was inserted into the dB in order to avoid compatibility issues?

[kriewall]

@maca88 I managed to confirm that the data inserted is slightly different when OracleUseBinaryFloatingPointTypes is on versus off ... surprisingly (to me, anyway), the precision seems to be increased when the setting is true.

OracleUseBinaryFloatingPointTypes = false:

OracleUseBinaryFloatingPointTypes = true

In either case, GetDecimal() is still being called internally (same stack trace as above). I agree with @hazzik that this seems to be a bug with the GetValue() method on the provider; but is there anything I can do to get NHibernate to call GetDouble() instead of GetValue()?

I did happen to come across this; the final response might provide a workaround, although his implementation isn't working out of the box for me, so I'll have to figure out how to enable it.

[maca88]

The registration that happens in Oracle10gDialect.RegisterFloatingPointTypeMappings() is applied at data intake, correct?

Yes, but is it is also used to create columns of type binary_double instead of DOUBLE PRECISION.

I tried it out but now I'm seeing InvalidCastExceptions

That is because you most likely didn't change the column type to binary_double. In case you cannot change the column type to binary_double, then it is better to not use it. I think that oracle driver decides which method to call depending on the column type of the value. The advantage of binary_double is that is equivalent to C# double as both complies with IEEE 754 standard, where for DOUBLE PRECISION the driver need to manually convert it double and as you pointer out, it is not always possible. So I would recommend you to change the columns by using a script in case you cannot recreate the database.

[kriewall]

@maca88 Thanks. I can look into changing the column types; the challenge is that I'm using Fluent Migrator to generate the dB, and it appears to be using double precision instead of binary_double to generate the columns. But that's a feature request for them. :)

I just came across this about the same time that I received the response below from Oracle support.

In ODP.NET 19.10, we added a new property to OracleDataReader and OracleDataAdapter, SuppressGetDecimalInvalidCastException, for this exact use case. Set the property value to true and it will round off the NUMBER value and suppress getting the exception.

Is this a configuration setting that could be added to NHibernate? It seems like this would solve my problem very easily.

[kriewall]

@maca88 Direct link to SuppressGetDecimalInvalidCastException documentation here in case it's useful.

[maca88]

Is this a configuration setting that could be added to NHibernate?

Currently NHibernate doesn't have any interceptor for DbDataReader where this property can be set and it is something that could be added in the future. As for now, the only option you is to extend OracleManagedDataClientDriver, override CreateCommand and GenerateCommand methods by calling the base method and wrapping the result into a custom command that wraps OracleCommand. In the custom command override ExecuteDbDataReaderAsync and ExecuteDbDataReader, where prior returning set the proeprty SuppressGetDecimalInvalidCastException to true. Finally register the extended diver instead of OracleManagedDataClientDriver in the NHibernate configuration. Example code:

public class WrappedCommand : DbCommand
{
  private OracleCommand _command;

  public WrappedCommand(OracleCommand command)
  {
    _command = command;
  }

  protected override async Task<DbDataReader> ExecuteDbDataReaderAsync(CommandBehavior behavior, CancellationToken cancellationToken)
  {
    var reader = (OracleDataReader) await _command.ExecuteReaderAsync(behavior, cancellationToken);
    reader.SuppressGetDecimalInvalidCastException = true;
    return reader;
  }

  protected override DbDataReader ExecuteDbDataReader(CommandBehavior behavior)
  {
    var reader = _command.ExecuteReader(behavior);
    reader.SuppressGetDecimalInvalidCastException = true;
    return reader;
  }

  // TODO: wrap missing methods
}

public class CustomDriver : OracleManagedDataClientDriver
{
  public override DbCommand CreateCommand()
  {
    return new WrappedCommand((OracleCommand) base.CreateCommand());
  }

  public override DbCommand GenerateCommand(CommandType type, SqlString sqlString, SqlType[] parameterTypes)
  {
    return new WrappedCommand((OracleCommand) base.GenerateCommand(type, sqlString, parameterTypes));
  }
}

[kriewall]

Got it - thanks for the tip! I enabled SuppressGetDecimalInvalidCastException in my custom query routine and it had absolutely no observable effect. I've sent a note to Oracle inquiring about it. I'll try out your method once I understand what either they or I are doing wrong. :)

[HakanKaraoglu]

@kriewall
Have you had any feedback regarding your problem? I'm having the similar problem too.
See #2914

[kriewall]

@HakanKaraoglu Sorry for the slow response. I did contact Oracle about it; they acknowledged it was a bug and committed to fix it in an upcoming version. However, I ended up putting in custom logic to limit the precision on values going to the dB since I couldn't wait for Oracle's release schedule and lost track of it after that. I would be curious to know whether a version of Oracle.ManagedDataAccess more recent than 19.10 resolves the issue, in case you find yourself checking it out.

[HakanKaraoglu]

hello @kriewall,
I had to find my own solution and go on :) that's what I did

[kriewall]

@maca88 I need to revive this conversation. I put in a local patch for the issue I faced a couple years ago, but I just ran into another problem pertaining to this behavior with the Oracle client. I've taken another look at the fix you recommended, but I'm continually running into InvalidCastExceptions, this time pertaining to the inability to cast between OracleCommand and WrappedCommand. I ended up with the following implementation since the incoming command from CreateCommand and GenerateCommand was not always of a consistent type:

public class CustomDriver : OracleManagedDataClientDriver
{
    public override DbCommand CreateCommand()
    {
        var command = base.CreateCommand();
        return command.GetType() == typeof(WrappedCommand) ? command : new WrappedCommand((OracleCommand)command);
    }

    public override DbCommand GenerateCommand(CommandType type, SqlString sqlString, SqlType[] parameterTypes)
    {
        var command = base.GenerateCommand(type, sqlString, parameterTypes);
        return command.GetType() == typeof(WrappedCommand) ? command : new WrappedCommand((OracleCommand)command);
    }
}

Once I did this, I started to see InvalidCastExceptions thrown by OracleDataClientDriverBase.OnBeforePrepare(DbCommand command). I attempted to resolve this by overriding OnBeforePrepare:

protected override void OnBeforePrepare(DbCommand command)
{
    var cmd = command.GetType() == typeof(WrappedCommand) ? ((WrappedCommand)command).OracleCommand : command;
    base.OnBeforePrepare(cmd);
}

... but that yields a NullReferenceException from the private method OracleDataClientBatchingBatcher.SetArrayBindCount(Int32 arraySize). I investigated a custom implementation of OracleDataClientBatchingBatcher, but it looks like that would require access to internal members within NHibernate.

So, I have a number of questions:

• Is there something else I need to override or some other straightforward custom implementation I need to get this to fall into place?
• I've tried specifying precision in the mapping, but it doesn't have any effect on writes to or reads from the database. If I set the precision very low, I should expect to see that reflected in the dB, yes? I am using Fluent NHibernate, so I've raised an issue with them here, but I'm not sure if my expectations are correct or if this is related to NHibernate itself.
• You mentioned above that adding an interceptor for DbDataReader such that SuppressGetDecimalInvalidCastException could be set is an option. Is this something that could be added to your roadmap? I'm hesitant to offer submitting a merge request myself since I might need a fair amount of guidance depending on the complexity and required knowledge of NHibernate internals

[fredericDelaporte]

I wonder if we should revert this change from v5.1. As written in the #1607 comments, it was done because:

Oracle: double raised to 40 by the way, because it stores them as decimal but read them as .Net double, so it should not be limited to .Net decimal capabilities.

But since the Oracle driver is quite bad at reading them back, this was likely bad move. (But those Oracle driver failures to read them bask is in my opinion a defect that ideally should be fixed on their side, but it seems to take quite some time to happen.)

So, reverting this v5.1 change may avoid having those "too high precision" non binary doubles inserted. Would you be able to check this with a custom dialect reverting that change, or by testing with NHibernate 5.0.8?

[fredericDelaporte]

As you are using Oracle 11, you can enable binary floating point types, by setting configuration setting oracle.use_binary_floating_point_types to true.

We have forgotten to update the html documentation about this new setting. If we revert the 5.1 change I am pinpointing above, we should think about documenting this setting here by the way.

[fredericDelaporte]

So, reverting this v5.1 change may avoid having those "too high precision" non binary doubles inserted.

On second thought, no, it will not change anything. It would only change the Oracle declared type when also setting in your mapping a precision on a double higher than the maximum set in the dialect.

[maca88]

Is there something else I need to override or some other straightforward custom implementation I need to get this to fall into place?

As far as I know, there is no other straightforward solution for this. You need to create a custom batcher that will unwrap the command before setting ArrayBindCount property in the SetArrayBindCount method.

I investigated a custom implementation of OracleDataClientBatchingBatcher, but it looks like that would require access to internal members within NHibernate.

As SetArrayBindCount method is private, you need to copy the whole OracleDataClientBatchingBatcher code and modify the setting of ArrayBindCount property. I know this is not ideal, but currently, it is the only way. There is already an open issue #1586 for this and when solved, it won't be required to create custom batchers anymore.

I've tried specifying precision in the mapping, but it doesn't have any effect on writes to or reads from the database.

Just by setting the mapping is not enough, you also need to change the column in the database in order it to work.

You mentioned above that adding an interceptor for DbDataReader such that SuppressGetDecimalInvalidCastException could be set is an option. Is this something that could be added to your roadmap?

I think that a simpler solution would be to add a setting (similar to oracle.use_binary_floating_point_types) that would call SuppressGetDecimalInvalidCastException, when enabled. As this should be a fairly simple change, I can try to make a pull request for it in the following days.

[kriewall]

Thank you so much for the speedy responses! A simple property setting would be wonderful. I will wait for your more elegant solution since this issue doesn't occur that often and I can manage it manually when it does. If something comes up and the feature will be delayed for more than, say, a month, please let me know and I'll rethink my decision.

[kriewall]

Hi @maca88 @fredericDelaporte I apologize but I've installed NHibernate 5.3.11 and I'm not seeing OracleSuppressDecimalInvalidCastException on the Environment class, although the changes are there are on the merge request above. I've loaded the DLL using ILSpy from my %userprofile%\.nuget\packages\NHibernate\5.3.11\lib\net461\NHibernate.dll and I don't see it there either. It doesn't appear to be included in the 5.3.11 release notes. Is it still pending release? If so, roughly when can I expect it on Nuget?

[fredericDelaporte]

That is not in 5.3.11, this is a "next minor" feature (likely 5.4), not yet published.
You can test it through MyGet nightly builds.

[kriewall]

Thank you. I installed the latest dev build from the nightly MyGet feed; however, it doesn't seem to resolve the issue. I've verified the following;

• The Nhibernate library in my bin/debug/net472 folder is version 5.4.0.3725
• The Oracle.ManagedDataAccess dll in my bin/debug/net472 folder is version 4.122.21.1:20211222, which matches the dll from my local Oracle.ManagedDataAccess/21.5.0/lib/net462 nuget cache.
• This line in my SessionFactory compiles successfully: x.SetProperty(Environment.OracleSuppressDecimalInvalidCastException, "true");
• The property oracle.suppress_decimal_invalid_cast_exception is set successfully if I inspect the NHibernate configuration after the previous command is reached.
  

This is the specific value that is causing the InvalidCastException: 0.0000000000000000711681426041767. I don't know if this matters, but my dB version is Oracle 11.2; but again my Oracle.ManagedDataAccess dll is version 21.5.0, so I wouldn't think the dB version would have anything to do with it. Am I missing something or is there something else I can check?

Thanks for your continued support. I've got to get this fixed or I've got a real mess on my hands. :|

[maca88]

I updated the test that was added for this issue by:

1. Changing the value to the one you mentioned
2. Updating the test project to net472 and Oracle.ManagedDataAccess to version 21.5.0
3. Adding OracleDataClientBatchingBatcherFactory for batching

and I was still not able to get the InvalidCastException. If you could create a simple test project or update the mentioned test so that it will throw the exception, would help to find and resolve the issue.

[kriewall]

Just a quick update: I downloaded the source code but first was getting a slew of ambiguous call errors on Assert.That; those didn't seem to pertain to the tests that you mentioned, so I just commented out the failing classes. Now, however, the tests are getting skipped for the following reason: OneTimeSetUp: NHibernate.Test.NHSpecificTest.GH2641.FixtureAsync does not apply to NHibernate.Dialect.MsSql2008Dialect. Is there some configuration I need to change somewhere?

One difference I note is that I'm using Oracle10gDialect. Just for starters I planned to switch OracleSuppressDecimalInvalidCastException to false and make sure that an exception actually does get thrown, but I'll have to get the unit test up and running first. Also, not sure if this matters (I didn't see anything pertaining to db/driver specification in your fixture), but this is my dB config:

                var dbConfig = OracleDataClientConfiguration.Oracle10
                    .ConnectionString(c => c.Is(connectionString))
                    .Driver<NHibernate.Driver.OracleManagedDataClientDriver>();

I have requested my counterpart to set up the dummy project you requested; hopefully we can get that to you soon.

[maca88]

Is there some configuration I need to change somewhere?

Yes, you need to modify the App.config in the NHibernate.Test project by replacing out the following lines:

<property name="dialect">NHibernate.Dialect.MsSql2008Dialect</property>
<property name="connection.driver_class">NHibernate.Driver.Sql2008ClientDriver</property> <!-- Shouldn't be necessary, but is required by some tests -->
<property name="connection.connection_string">Server=localhost\sqlexpress;Database=nhibernate;Integrated Security=SSPI</property>

with:

<property name="dialect">NHibernate.Dialect.Oracle10gDialect</property>
<property name="connection.driver_class">NHibernate.Driver.OracleManagedDataClientDriver</property>
<property name="oracle.use_n_prefixed_types_for_unicode">true</property>
<property name="oracle.use_binary_floating_point_types">false</property>
<property name="connection.connection_string">User ID=nhibernate;Password=nhibernate;Data Source=localhost:59261/XE;Metadata Pooling=false;Self Tuning=false;</property>

and change the connection.connection_string to match your local oracle instance.

[kriewall]

@maca88 Okay, your code is fine; there was a developer error on our side. I've verified that the InvalidCastExceptions are getting suppressed in our application.

Thanks again for taking the time to support this request - it's a huge relief to have this issue under control.

[fredericDelaporte]

About running the tests, as explained in contributing, use the build menu. It helps configuring your database. Once configured, you can run the tests with whatever NUnit compatible tool you like, including VS Test Explorer. (You are not restricted to using the build menu for running the tests.)

[kriewall]

@fredericDelaporte Thanks for the guidance, I'll make note of it for the next time. I was able to get the unit tests of interest running using @maca88's info; they do indeed pass when the suppression flag is enabled and fail for InvalidCastException when disabled.