MassTransit Worker registration AFTER creating the bus - masstransit

So I've just pulled the latest MT from NuGet and have been coding with it like a bandit. Love it so far. Now the scenario I am currently implementing is a generic "bus" wrapper that one of the implementations is MT. So in the concrete implementation of the MT I implement the Bus in the constructor... lets call it 'register'
/// <summary>
/// Registers this instance with a particular bus. Every instance with the same name
/// will end up on the same bus
/// </summary>
/// <param name="networkName">The common key to share with other instances on the network</param>
public void Register(string networkName)
{
_bus = ServiceBusFactory.New(x =>
{
x.ReceiveFrom("msmq://localhost/the_wheels_on_the_bus");
x.SetPurgeOnStartup(true);
x.SetNetwork(networkName);
x.UseMsmq();
x.UseMulticastSubscriptionClient();
});
}
Now, abstract members allow the external code to register handlers using this implementation
/// <summary>
/// Abstract method to implement on the bus for allow for Responding to Request/Response calls
/// </summary>
/// <typeparam name="TIn">The message to be received from the requester.<remarks>Implements <see cref="ICorrelatedMessage"/>ICorrelatedMessage</remarks> and must have a value</typeparam>
/// <typeparam name="TOut">The message to be returned to the requester.<remarks>Implements <see cref="ICorrelatedMessage"/>ICorrelatedMessage</remarks> and must have a value</typeparam>
/// <param name="hostedClassesFunc">The func to invoke to get the appropriate data</param>
protected override void RegisterHandler<TIn, TOut>(Func<TIn, TOut> hostedClassesFunc)
{
_bus.SubscribeHandler<TIn>(msg =>
{
var output = hostedClassesFunc.Invoke(msg);
var context = _bus.MessageContext<TIn>();
context.Respond(output);
});
if (typeof(TIn).GetInterfaces().Contains(typeof(ICachableItem))
{
//mark it as a worker
//_bus.Worker ?? <-- How can i register this guy here?
}
}
So the question lies here, Can i even register a worker here? I cant seem to find anywhere to inject a worker on the bus at this point.
Thanks in advance for the help.. i hop the code ends up looking right

_bus.SubscribeConsumer<SubscriberType>() or _bus.SubscribeConsumer(instanceOfSubscriberType) will subscribe normal consumers. If you want distributed worker consumers, you must do it inside the configuration block - we have not exposed the option to do so on the bus itself.
However, it's really important that you subscribe consumers inside the ServiceBusFactory.New block. Otherwise MT will start consuming messages on the queue and if there are no consumers registered for that type because you haven't done it yet, messages will end up in the error queue.
I don't see any other code, so this might not be a problem but remember each IServiceBus needs unique a RecieveFrom.

There currently isnt any way to do this outside of the configuration block. This would have to be exposed somehow in order to do it after the bus is created.

Related

MvvmCross - navigation with custom objects

I have followed the steps in this link
Passing complex navigation parameters with MvvmCross ShowViewModel
i implemented an instance of the IMvxJsonConverter, and registered it. this is my code for my view model
public class AccountDetailsViewModel : BaseViewModel<AccountDetailsNav>
{
private readonly Repository.AccountsRepository _accounts;
Account _fullAccount;
public AccountDetailsViewModel(Repository.AccountsRepository accounts)
{
_accounts = accounts;
}
protected override void RealInit(AccountDetailsNav parameter)
{
//stuff
}
I have tried simple types by just passing thru strings , this is the code i use to navigate to to the viewmodel
Mvx.RegisterSingleton<Repository.AccountsRepository>(() =>
{
return _accounts;
});
ShowViewModel<AccountDetailsViewModel>(nav);
But it never ever seems to arrive in my view model methods or populates my data, and i cannot for the life of me figure out why. the data is serialized fine , and i have even tried blank constructors to no avail .. i just cannot figure out why its not hitting the realinit
K i found the problem , when adding a new view i failed to remove this method on the code behind of the view, and as such was causing my viewmodel to be null and never hitting my breakpoints
/// <summary>
/// Invoked when this page is about to be displayed in a Frame.
/// </summary>
/// <param name="e">Event data that describes how this page was reached.
/// This parameter is typically used to configure the page.</param>
protected override void OnNavigatedTo(NavigationEventArgs e)
{
}

Uninitialised JsonSerializer in Breeze SaveBundleToSaveMap sample

I'm attempting to use the SaveBundleToSaveMap snippet linked below to implement custom save handling on the server side of a breeze web api implementation.
SaveBundleToSaveMap
This sample does not work as is? (see below); their is a null reference exception which could use some attention.
The SaveWorkState(provider, entitiesArray) constructor calls the ContextProvider.CreateEntityInfoFromJson(...) method which then calls (the class scoped) JsonSerializer.Deserialize(new JTokenReader(jo), entityType) method.
The issue is that JsonSerializer is uninitialised and we get a null reference exeption.
For e.g. I added this test hack to get the code running:
protected internal EntityInfo CreateEntityInfoFromJson(dynamic jo, Type entityType) {
//temp fix to init JsonSerializer if SaveChanges has NOT been called
if(JsonSerializer==null) JsonSerializer = CreateJsonSerializer();
var entityInfo = CreateEntityInfo();
entityInfo.Entity = JsonSerializer.Deserialize(new JTokenReader(jo), entityType);
entityInfo.EntityState = (EntityState)Enum.Parse(typeof(EntityState), (String)jo.entityAspect.entityState);
entityInfo.ContextProvider = this;
This issue does not occur in the standard release bits as CreateEntityInfoFromJson is always? called downstream from a SaveChanges() call which means the JsonSerializer gets initialised.
However, things would be better structured if an initialised JsonSerializer was passed to CreateEntityInfoFromJson as a parameter to avoid potential future null reference issues?
Alternately, is there a way to get the SaveBundleToSaveMap snippet to init the JsonSerializer? Its got a private setter :(
UPDATE
Implemented a very hacky stopgap solution. If anyone at IdeaBlade is watching, would be great to have a public API to convert to and from json saveBundle <-> saveMap.
/// <summary>
/// Convert a json saveBundle into a breeze SaveMap
/// </summary>`enter code here`
public static Dictionary<Type, List<EntityInfo>> SaveBundleToSaveMap(JObject saveBundle)
{
var _dynSaveBundle = (dynamic)saveBundle;
var _entitiesArray = (JArray)_dynSaveBundle.entities;
var _provider = new BreezeAdapter();
//Hack 1: Breeze.ContextProvider initializes a global JsonSerializer in its SaveChanges() method
//We are bypassing SaveChanges() and bootstrapping directly into SaveWorkState logic to generate our saveMap
//as such we need to init a serializer here and slipsteam it in via reflection (its got a private setter)
var _serializerSettings = BreezeConfig.Instance.GetJsonSerializerSettings();
var _bootstrappedJsonSerializer = JsonSerializer.Create(_serializerSettings);
//Hack 2:
//How to write to a private setter via reflection
//http://stackoverflow.com/questions/3529270/how-can-a-private-member-accessable-in-derived-class-in-c
PropertyInfo _jsonSerializerProperty = _provider.GetType().GetProperty("JsonSerializer", BindingFlags.Instance | BindingFlags.NonPublic);
//Hack 3: JsonSerializer property is on Breeze.ContextProvider type; not our derived EFContextProvider type so...
_jsonSerializerProperty = _jsonSerializerProperty.DeclaringType.GetProperty("JsonSerializer", BindingFlags.Instance | BindingFlags.NonPublic);
//Finally, we can init the JsonSerializer
_jsonSerializerProperty.SetValue(_provider, _bootstrappedJsonSerializer);
//saveWorkState constructor loads json entitiesArray into saveWorkState.EntityInfoGroups struct
var _saveWorkState = new SaveWorkState(_provider, _entitiesArray);
//BeforeSave logic loads saveWorkState.EntityInfoGroups metadata into saveWorkState.SaveMap
_saveWorkState.BeforeSave();
var _saveMap = _saveWorkState.SaveMap;
return _saveMap;
}
I looked into this. You don't actually need to make a change to the Breeze code to accomplish what you want. The ContextProvider is designed such that you can do just about whatever you want during save.
I'm curious: what "custom save handling" do you want to perform that you can't do today with the BeforeSave and AfterSave logic? I see in your "stopgap" code that you're calling BeforeSave on the SaveWorkState. What more do you need?
As an exercise, I wrote a NorthwindIBDoNotSaveContext that does what you want. Here's how it goes:
/// <summary>
/// A context whose SaveChanges method does not save
/// but it will prepare its <see cref="SaveWorkState"/> (with SaveMap)
/// so developers can do what they please with the same information.
/// See the <see cref="GetSaveMapFromSaveBundle"/> method;
/// </summary>
public class NorthwindIBDoNotSaveContext : EFContextProvider<NorthwindIBContext_CF>
{
/// <summary>
/// Open whatever is the "connection" to the "database" where you store entity data.
/// This implementation does nothing.
/// </summary>
protected override void OpenDbConnection(){}
/// <summary>
/// Perform your custom save to wherever you store entity data.
/// This implementation does nothing.
/// </summary>
protected override void SaveChangesCore(SaveWorkState saveWorkState) {}
/// <summary>
/// Return the SaveMap that Breeze prepares
/// while performing <see cref="ContextProvider.SaveChanges"/>.
/// </summary>
/// <remarks>
/// Calls SaveChanges which internally creates a <see cref="SaveWorkState"/>
/// from the <see param="saveBundle"/> and then runs the BeforeSave and AfterSave logic (if any).
/// <para>
/// While this works, it is hacky if all you want is the SaveMap.
/// The real purpose of this context is to demonstrate how to
/// pare down a ContextProvider, benefit from the breeze save pre/post processing,
/// and then do your own save inside the <see cref="SaveChangesCore"/>.
/// </para>
/// </remarks>
/// <returns>
/// Returns the <see cref="SaveWorkState.SaveMap"/>.
/// </returns>
public Dictionary<Type, List<EntityInfo>> GetSaveMapFromSaveBundle(JObject saveBundle)
{
SaveChanges(saveBundle); // creates the SaveWorkState and SaveMap as a side-effect
return SaveWorkState.SaveMap;
}
}
And here's how you could use it to get the SaveMap:
var saveMap = new NorthwindIBDoNotSaveContext().GetSaveMapFromSaveBundle(saveBundle);
Yes, it is "hacky", particularly if all you want is the SaveMap. But why do you just want the SaveMap?
We've designed the ContextProvider (and all of its sub-classes) such that you have free reign over the SaveChangesCore method. You could override that, further manipulate the SaveMap, then either delegate to the base implementation or do whatever else you have in mind for saving the entity data.
But while I don't see what you're after, it was not all that hard to extract the SaveChanges initialization logic into its own method.
So in the next release (after 1.5.2), you should find the following new method in the ContextProvider:
protected void InitializeSaveState(JObject saveBundle)
{
JsonSerializer = CreateJsonSerializer();
var dynSaveBundle = (dynamic)saveBundle;
var entitiesArray = (JArray)dynSaveBundle.entities;
var dynSaveOptions = dynSaveBundle.saveOptions;
SaveOptions = (SaveOptions)JsonSerializer.Deserialize(new JTokenReader(dynSaveOptions), typeof(SaveOptions));
SaveWorkState = new SaveWorkState(this, entitiesArray);
}
SaveChanges now calls that method before continuing on in its previous manner:
public SaveResult SaveChanges(JObject saveBundle, TransactionSettings transactionSettings = null) {
if (SaveWorkState == null || SaveWorkState.WasUsed) {
InitializeSaveState(saveBundle);
}
transactionSettings = transactionSettings ?? BreezeConfig.Instance.GetTransactionSettings();
...
}
Notice that SaveChanges won't call InitializeSaveState twice if you've already prepared the SaveWorkState by, say, calling InitializeSaveState externally and then called SaveChanges immediately thereafter. It also won't save twice with a "used" SaveWorkState.
The source is checked into github right now if you're interested.
You'll be able to get the SaveMap from a save bundle by adding this method to your sub-class of a ContextProvider as in this example:
public class NorthwindContextProvider: EFContextProvider<NorthwindIBContext_CF> {
...
public Dictionary<Type, List<EntityInfo>> GetSaveMapFromSaveBundle(JObject saveBundle) {
InitializeSaveState(saveBundle); // Sets initial EntityInfos
SaveWorkState.BeforeSave(); // Creates the SaveMap as byproduct of BeforeSave logic
return SaveWorkState.SaveMap;
}
...
}
Now you use that as follows:
var saveMap = ContextProvider.GetSaveMapFromSaveBundle(saveBundle);

Location aware code converted from C# to F#?

I have the following code in a WP7 app, and am starting to look at F#.. I can't find any GeoCoordinate examples, can anyone give me an idea of how this code would look in F#? Or point me to an example? I've had a look at some tutorials, books and Pluralsight, so think I am just starting to grasp the basics..but can't seem to get my head around this! All the examples I can seem to find are based around mathematical problem spaces. Any help or advice is much appreciated!
public partial class MainPage : PhoneApplicationPage
{
GeoCoordinateWatcher watcher;
// Constructor
public MainPage()
{
InitializeComponent();
SupportedOrientations = SupportedPageOrientation.Portrait | SupportedPageOrientation.Landscape;
}
private void btnStart_Click(object sender, RoutedEventArgs e)
{
//Reinitialize the GeoCoordinateWatcher
watcher = new GeoCoordinateWatcher(GeoPositionAccuracy.High);
watcher.MovementThreshold = 100;//distance in meters
//Add event handlers for StatusChanged and PositionChanged Events
watcher.StatusChanged += new EventHandler<GeoPositionStatusChangedEventArgs>(watcher_StatusChanged);
watcher.PositionChanged += new EventHandler<GeoPositionChangedEventArgs<GeoCoordinate>>(watcher_PositionChanged);
//Start data acquisition
watcher.Start();
//hide button
btnStart.Visibility = Visibility.Collapsed;
}
#region Event Handlers
/// <summary>
/// Handler for the StatusChanged event. This invokes MyStatusChanged on the UI thread
/// and passes the GeoPositionStatusChangedEventArgs
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
void watcher_StatusChanged(object sender, GeoPositionStatusChangedEventArgs e)
{
Deployment.Current.Dispatcher.BeginInvoke(() => MyStatusChanged(e));
}
/// <summary>
/// Handler for the PositionChanged Event. This invokes MyPositionChanged on the UI thread and
/// passes the GeoPositionStatusChangedEventArgs
/// </summary>
/// <param name="sender"></param>
/// <param name="e"></param>
void watcher_PositionChanged(object sender, GeoPositionChangedEventArgs<GeoCoordinate> e)
{
Deployment.Current.Dispatcher.BeginInvoke(() => MyPositionChanged(e));
}
#endregion
/// <summary>
/// Custom method called from the PositionChanged event handler
/// </summary>
/// <param name="e"></param>
/// <returns></returns>
void MyPositionChanged(GeoPositionChangedEventArgs<GeoCoordinate> e)
{
//update the map to show the current location
GeoCoordinate geo = new GeoCoordinate(e.Position.Location.Latitude, e.Position.Location.Longitude);
Location ppLoc = new Location(e.Position.Location.Latitude, e.Position.Location.Longitude);
mapMain.SetView(geo, 10);
//update pushpin location and show
MapLayer.SetPosition(ppLocation, ppLoc);
ppLocation.Visibility = System.Windows.Visibility.Visible;
}
/// <summary>
/// Custom method called from the StatusChanged event handler
/// </summary>
/// <param name="e"></param>
/// <returns></returns>
void MyStatusChanged(GeoPositionStatusChangedEventArgs e)
{
switch (e.Status)
{
case GeoPositionStatus.Disabled:
//the location service is disabled or unsupported, alert the user
tbStatus.Text = "Sorry we can't find you on this device";
break;
case GeoPositionStatus.Initializing:
//location service is initializing
//disable the start location button
tbStatus.Text = "Looking For you...";
break;
case GeoPositionStatus.NoData:
//location service is working but no data found, alert the user and enable the stop location button
tbStatus.Text = "can't find you yet...";
ResetMap();
break;
case GeoPositionStatus.Ready:
//location service is receiving data, show the current position and enable the stop location button
tbStatus.Text = "We found you!";
break;
}
}
void ResetMap()
{
Location ppLoc = new Location(0, 0);
GeoCoordinate goe = new GeoCoordinate(0.0,0.0);
mapMain.SetView(goe, 1);
//update pushpin location and show
MapLayer.SetPosition(ppLocation, ppLoc);
ppLocation.Visibility = System.Windows.Visibility.Collapsed;
}
}
I think that this is due to the fact that F# is touted as a language that you can process a large amount of information without being very verbose. While you can build small user interface elements using F# by calling relevant libraries, the intention is for you to build UI's with C# / ASP.NET/ etc. So, it wouldn't really make sense with your application because all you are doing is building a small UI and connecting events of that UI to a larger library of geoprocessing capabilities.
But if you wanted to collect information from that library (or a similar one) of all the points of interest nearby, then sort them according to distance from the user and his potential for 'liking' that point of interest based on some algorithm designed to compare a random point of interest with catagories or prior ratings then F# would be a good choice. You can rapidly describe those data structures, manipulate them, and return the result of it's processing back to your user interface.
This is why instructions such as the one shown here can be helpful. While very light on calling or creating a user interface (the C# code just display some text passed from the F# code), it can be used to create a backend for your phone application.
The sample you posted is a lot of code, so I don't expect that anybody will translate that to F# for you. Calling .NET functionality from F# is generally quite similar to how you'd call it from C# (at least initially, before you learn how to use some advanced F# patterns), so the translation should be pretty direct.
The F# version of code that initializes the GeoCoordinateWatcher is probably going to look like this:
let watcher = new GeoCoordinateWatcher(GeoPositionAccuracy.High)
watcher.MovementThreshold <- 100
// Add event handlers for StatusChanged and PositionChanged Events
watcher.StatusChanged.Add(fun eargs ->
MyStatusChanged(eargs) )
watcher.PositionChanged.Add(fun eargs ->
MyPositionChanged(eargs) )
// Start data acquisition
watcher.Start()
In general, F# has a couple of nice features that simplify user interface programming. As far as I know, there isn't a guide on developing Windows Phone applications in F#, specifically, but MSDN has a section that describes development of Silverlight applications, and most of the patterns will be the same:
Developing Client-Side Applications - Real World Functional programming on MSDN

Bind a Global Action Filter to all Controllers in an Area using MVC 3 Dependency Injection with Ninject 2.2

I was able to to use ASP.NET MVC 3 and Ninject 2.2 to inject a logger object into a custom ActionFilterAttribute thanks to the help I received in this post.
Now I would like to bind my custom ActionFilterAttribute only to all controllers that are in a specific area.
I was able to get started with the following binding but it only handles one controller in a certain area. I would like my code to bind to all controllers in a specific area. Any ideas?
/// <summary>
/// Load your modules or register your services here!
/// </summary>
/// <param name="kernel">The kernel.</param>
private static void RegisterServices(IKernel kernel)
{
kernel.Bind<ILogger>().To<Log4NetLogger>().InRequestScope();
kernel.BindFilter<TestLoggingAttribute>(
FilterScope.Controller, 0)
.WhenControllerType<OrganizationController>();
}
This helped me, Thanks Darin. However, context.RouteData.Values did not have the area for me but context.RouteData.DataTokens["area"] did! also in my case I had controller that were not in specific areas (e.g. shared controllers) therefore I had to check for datatoken area to be null. This is what worked for me:
kernel
.BindFilter<TestLoggingAttribute>(FilterScope.Controller, 0)
.When((context, ad) => context.RouteData.DataTokens["area"] != null && context.RouteData.DataTokens["area"] == "Organization");
kernel
.BindFilter<TestLoggingAttribute>(FilterScope.Controller, 0)
.When((context, ad) => context.RouteData.Values["area"] == "YourAreaName");
or:
kernel
.BindFilter<TestLoggingAttribute>(FilterScope.Controller, 0)
.When((context, ad) => context.Controller.GetType().FullName.Contains("Areas.YourAreaName"));

How does "see" work in the method docs?

/// <summary>
/// Get all following siblings of each element up to but not including the element matched by the selector.
/// </summary>
/// <param name="selector">A string containing a selector expression to indicate where to stop matching following sibling elements.</param>
/// <see cref="http://api.jquery.com/nextUntil/"/>
/// <returns></returns>
public SharpQuery NextUntil(string selector = null)
{
throw new NotImplementedException();
}
I wanted to add a link in my method docs to link to a fuller explanation. "see" seemed appropriate for this (intellisense suggested it). However, when I call start typing my method, "see" doesn't appear in the tooltip. Is there a way to go to that link? I tried pressing F1, it took me to MSDN instead.
The <see> tag must be used within the text of other comment tags in order to specify a hyperlink.
You can also use <seealso> to specify a hyperlink to appear in a See Also section of the generated documentation.
MSDN provides the following example:
/// text for class TestClass
public class TestClass
{
/// <summary>DoWork is a method in the TestClass class.
/// <para>Here's how you could make a second paragraph in a description. <see cref="System.Console.WriteLine(System.String)"/> for information about output statements.</para>
/// <seealso cref="TestClass.Main"/>
/// </summary>
public static void DoWork(int Int1)
{
}
/// text for Main
static void Main()
{
}
}
From what I gathered at:
http://msdn.microsoft.com/en-us/library/5ast78ax(VS.80).aspx
This tags ( , ) will be available in the generated documentation file (the XML file, when you do /doc compiler options), and then further processed by tool like Sandcastle

Resources