VoltDB csharp wireprotocol and client implementation
Failed to load latest commit information.
ReleaseBuild Add ReleaseBuild directory to git repo so that the .sln works for rel… Mar 21, 2012
Help.shfbproj update version, readme and build instructions Aug 5, 2016
VoltDB.Examples.VS2008.sln Reorganized for 1.3.2 May 2, 2011
VoltDB.Examples.VS2012.sln Visual Studio 2012 solution files Mar 5, 2013
build.documentation.txt update version, readme and build instructions Aug 5, 2016


VoltDB C# Client Library

The VoltDB client library implements the native VoltDB wire
protocol. You can use the library to connect to a VoltDB cluster,
invoke stored procedures and read responses.

For an introduction to using the C# library:
[Introducting VoltDB.NET: C# Library for Your VoltDB Applications](https://voltdb.com/blog/introducing-voltdbnet-c-library-your-voltdb-applications)

New Features in V6.0
Fixes for VoltDecimal where some of the Decimal values were transmitted incorrectly over the wire.

## Getting Started ##

Download the driver:
- Latest build with documentation: [Downloads](https://downloads.voltdb.com/loader.php?kit=CsharpClient)
- Full source on Github: [voltdb-client-csharp](https://github.com/VoltDB/voltdb-client-csharp)

## VoltDB.NET in a nutshell ##

The .NET/C# client library for VoltDB is extremely flexible and allows you to develop WinForms, Console and Web applications and services much as you would leveraging any other back-end data service or database.

- Connect with a customized connectionstring in your app/web.config.
- Perform Synchronous or Asynchronous data operations using standard .NET design patterns.
- Consume and filter data results using LINQ.
- Leverage native, strongly-typed, data access and type casting, as well as late-bound operations.
- Monitor performance and manage a database cluster.

Key considerations:

- All access is thread-safe
    - Safe for multi-threaded application
    - Safe as shared connections for a website/service
    - OK to mix asynchronous and synchronous calls
    - OK to have long-running callbacks (they are Thread-Pooled)
- Resilience
    - Connect to multiple nodes
- Performance
    - Share: multi-threading on a single connection is faster
    - Production: No Tracing (ever) or Statistics (unless needed)
    - Avoid usage of IAsync WaitHandles: they are slow!

## Connecting to VoltDB ##

Read the section "Creating a Database Connection" of blog post: [Introducting VoltDB.NET: C# Library for Your VoltDB Applications](https://voltdb.com/blog/introducing-voltdbnet-c-library-your-voltdb-applications).

Key Connection Options:
- Port – default: 21212
- MaxOutstandingTransaction (txnq) – default: 3,000
- Statistics (stats) – default: false
- CommandTimeout – default: 5,000(ms)
- AllowSystemCalls (system) – default: false
- AllowAdHocQueries (adhoc) – default: false

## Synchronous and Asynchronous Calls ##

Synchronous = Do nothing until you get the answer
- Good: ‘Procedural’ model easier to understand/work with
- Bad: Limits your throughput
- Redeeming: For this thread only

Asynchronous = Call me when you get the answer
- Good: No bottlenecks on wait operations
- Bad: ‘Disjointed’ model harder to work with/error prone
- Redeeming: Known ‘Best Practice’ patterns

VoltDB.NET implements:
- Sync: .Execute
- Async: .Begin/.Cancel/.End with IAsyncResult

# Example Code #

## Calling Procedures ##

Define Callback for Asynchronous call

    void MyDelegate(Response<Table[]> response) {
      if (response.Status == ResponseStatus.Success) {
        … // Send response to client, update UI, etc.
      else {
        … // Deal with error

Create Strongly-typed Wrappers

    var my = conn.Procedures.Wrap<Table[],int,string>("MyProcedure", MyDelegate);

Supported Data Types:

* Result:
    - Table[], Table
    - SingleRowTable[], SingleRowTable
    - int[], long[], string[], double[], … (and nullable types)
    - int, long, … (and nullable types)
* Parameters:
    - int[], long[], string[], double[], …
    - int, long, …

Execute the procedure

    Response<Table[]> r = myProc.Execute(1, “test”);

    IAsyncResult h = my.BeginExecute(1, “test”);
    IAsyncResult h = my.BeginExecute(1, “test”, state);

Cancel Async Execution


Get Async results (if not using callback)

    Response<Table[]> r = myProc.EndExecute(h);

Wrapper Rules:

- Up to 35 input parameters
- Types must be compatible with core VoltDB types
- Types can be single-values or arrays
    - sbyte, short, int, long, double, VoltDecimal, DateTime, string
    - sbyte?, short?, int?, …
    - sbyte[], sbyte?[], short[], short?[], …

Another option is to use Runtime Wrappers, but the type of runtime values must still be compatible.


Per-execution callback delegate/closures

    IAsyncResult h = my.BeginExecute(1, "test", MyDelegate);
    // or...
    h = my.BeginExecute( 1, "test", (r) => MyClosureFunction(r, …) );

Re-use Wrappers across connections (of course, executions still occur in the initiating context/connection).


## Consuming Results ##

Access data directly...

    double? value = response.Result
                            .GetValue<double?>(col, row);

    double? Value = response.Result

...or through Strongly-typed Table Wrappers

    var myTable = response.Result.Wrap<int?,…,double?>();

    double? value = myTable.Column7[row];

Wrapper Rules:

- Up to 35 columns
- Types must be compatible with core VoltDB types and flagged as Nullable => use int? (not int)
    - sbyte?, short?, int?, long?, double?, VoltDecimal?, DateTime? and string

Results are LINQ-friendly:

    // On a strongly-typed VoltDB data table
    myTable.Rows.Where(r => r.Column2 == "Books")
                .Select(r => new { Title = r.Column2, Price = r.Column7 })
                .OrderBy(p => p.Price);

    // On a raw VoltDB data table
    raw.Rows.Where(r => r.GetValue<string>(1) == "Books")
            .Select(r => new { Title = r. GetValue<string>(1), Price = r. GetValue<double>(6) })
            .OrderBy(p => p.Price);

Access metadata:

    int count = myTable.RowCount;
    bool check = myTable.HasData;

    string name = raw.GetColumnName(idx);
    short idx = raw.GetColumnIndex(name);

    Type type = raw.GetColumnType(idx);
    DBType type = raw.GetColumnDBType(idx);

    int?[] column1Data = raw.GetColumnData<int?>(0);
    object[] column1Data = raw.GetColumnData(0);

Fill a (System.Data.)DataTable

    Table raw = procedureWrapper.Execute().Result;
    DataTable dt = new DataTable("Result");
    for(short i = 0; i < raw.ColumnCount; i++)
        dt.Columns.Add( raw.GetColumnName(i), raw.GetColumnType(i));
    object[] values = new object[raw.ColumnCount];
    foreach (Row row in raw.Rows) {
        for (short i = 0; i < raw.ColumnCount; i++)
            values[i] = row.GetValue(i);

Fill a (System.Windows.Forms.)DataGridView

    view.DataSource = null;
    for (short i = 0; i &lt; raw.ColumnCount; i++)
        view.Columns.Add(raw.GetColumnName(i), raw.GetColumnName(i));
    foreach (Row row in raw.Rows) {
        int n = view.Rows.Add();
        for (short i = 0; i < row.ColumnCount; i++) {
            view.Rows[n].Cells[i].Value = row.GetValue(i);
            view.Rows[n].Cells[i].ValueType = row.GetColumnType(i);