Say you have a factory that returns instances of ILightBulb. Two ways (there may be more) of implementing the factory are as follows:
Option 1 - Passing in an enum type
enum LightBulbType
{
Incandescent,
Halogen,
Led,
}
class ILightBulbFactory
{
public ILightBulb Create(LightBulbType type)
{
switch (type)
{
case LightBulbType.Incandescent:
return new IncandescentBulb();
case LightBulbType.Halogen:
return new HalogenBulb();
case LightBulbType.Led:
return new LedBulb();
}
}
}
Option 2 - Explicit method names
class ILightBulbFactory
{
public ILightBulb CreateIncandescent()
{
return new IncandescentBulb();
}
public ILightBulb CreateHalogen()
{
return new HalogenBulb();
}
public ILightBulb CreateLed()
{
return new LedBulb();
}
}
Which method is most preferable, and why?
Thanks.
In java, you can solve it in the enum, thus making sure that if you add new enums, all your code will remain working, instead of forgetting to add statements to your case.
enum LightBulbType
{
Incandescent{
#Override
public ILightBulb getInstance() {
return new IncandescentBulb();
}
},
Halogen{
#Override
public ILightBulb getInstance() {
return new HalogenBulb();
}
},
Led{
#Override
public ILightBulb getInstance() {
return new LedBulb();
}
};
public abstract ILightBulb getInstance();
}
class ILightBulbFactory
{
public ILightBulb Create(LightBulbType type)
{
return type.getInstance();
}
}
The equivalent for c# would be
public sealed class LightBulbType
{
public static readonly LightBulbType Incandescent = new
LightBulbType("Incandescent", new IncandescentBulb());
public static readonly LightBulbType Halogen = new
LightBulbType("Halogen", new HalogenBulb());
public static readonly LightBulbType LedBulb = new LightBulbType("Led",
new LedBulb());
public static IEnumerable<LightBulbType> Values
{
get
{
yield return Incandescent;
yield return Halogen;
yield return LedBulb;
}
}
private string Name { get; set; }
private ILightBulb Instance { get;}
private LightBulbType(string name, ILightBulb instance)
{
this.Name = name;
this.Instance = instance;
}
public override string ToString() => Name;
public ILightBulb GetInstance()
{
return Instance;
}
}
public interface ILightBulb
{
}
public class IncandescentBulb : ILightBulb
{
}
public class HalogenBulb : ILightBulb
{
}
public class LedBulb : ILightBulb
{
}
public static class LightBulbFactory
{
public static ILightBulb Create(LightBulbType type)
{
return type.GetInstance();
}
}
For call
ILightBulb halogenBulb = new HalogenBulb();
ILightBulb incandescentBulb = new IncandescentBulb();
ILightBulb ledBulb = new LedBulb();
or
ILightBulb halo = LightBulbFactory.Create(LightBulbType.Halogen);
ILightBulb inca = LightBulbFactory.Create(LightBulbType.Incandescent);
ILightBulb led = LightBulbFactory.Create(LightBulbType.LedBulb);
Related
I tried to write a udf function to calculate my data. In the trino's docs, I knew I should to write a function plugin and I succeed to execute my udf aggregate function sql.
But when I write sql with aggregate function and window function, the sql executed failed.
The error log is com.google.common.util.concurrent.ExecutionError: java.lang.NoClassDefFoundError: com/example/ListState.
I think I may implement the interface about the window function.
The ListState.java file code
#AccumulatorStateMetadata(stateSerializerClass = ListStateSerializer.class, stateFactoryClass = ListStateFactory.class)
public interface ListState extends AccumulatorState {
List<String> getList();
void setList(List<String> value);
}
The ListStateSerializer file code
public class ListStateSerializer implements AccumulatorStateSerializer<ListState>
{
#Override
public Type getSerializedType() {
return VARCHAR;
}
#Override
public void serialize(ListState state, BlockBuilder out) {
if (state.getList() == null) {
out.appendNull();
return;
}
String value = String.join(",", state.getList());
VARCHAR.writeSlice(out, Slices.utf8Slice(value));
}
#Override
public void deserialize(Block block, int index, ListState state) {
String value = VARCHAR.getSlice(block, index).toStringUtf8();
List<String> list = Arrays.asList(value.split(","));
state.setList(list);
}
}
The ListStateFactory file code
public class ListStateFactory implements AccumulatorStateFactory<ListState> {
public static final class SingleListState implements ListState {
private List<String> list = new ArrayList<>();
#Override
public List<String> getList() {
return list;
}
#Override
public void setList(List<String> value) {
list = value;
}
#Override
public long getEstimatedSize() {
if (list == null) {
return 0;
}
return list.size();
}
}
public static class GroupedListState implements GroupedAccumulatorState, ListState {
private final ObjectBigArray<List<String>> container = new ObjectBigArray<>();
private long groupId;
#Override
public List<String> getList() {
return container.get(groupId);
}
#Override
public void setList(List<String> value) {
container.set(groupId, value);
}
#Override
public void setGroupId(long groupId) {
this.groupId = groupId;
if (this.getList() == null) {
this.setList(new ArrayList<String>());
}
}
#Override
public void ensureCapacity(long size) {
container.ensureCapacity(size);
}
#Override
public long getEstimatedSize() {
return container.sizeOf();
}
}
#Override
public ListState createSingleState() {
return new SingleListState();
}
#Override
public ListState createGroupedState() {
return new GroupedListState();
}
}
Thanks for help!!!!
And I found the WindowAccumulator class in the trino source code. But I don't know how to use it.
How to create a aggregate function for window function?
I have such a simple example:
public class Base<T> {
private String value;
public Base(String value) {
this.value = value;
}
#PostConstruct
private void init() {
System.out.println(value);
}
}
public class StringTest extends Base<String> {
public StringTest(String value) {
super(value);
}
}
public class IntegerTest extends Base<Integer> {
public IntegerTest(String value) {
super(value);
}
}
#Configuration
public class Configuration {
#bean(name="stringtest1")
public Base<?> getStringTest1() {
return new StringTest("string test 1");
}
#bean(name="stringtest2")
public Base<?> getStringTest2() {
return new StringTest("string test 2");
}
#bean(name="integertest1")
public Base<?> getIntegerTest1() {
return new IntegerTest("integer test 1");
}
#bean(name="integertest2")
public Base<?> getIntegerTest2() {
return new IntegerTest("integer test 2");
}
}
The output is
string test 1
string test 2
PostConstruct is only invoked for beans stringtest1 and stringtest2 but not for integertest1 and integertest2.
If I change the configuration to below (change ? to the concrete type)
#bean(name="stringtest1")
public Base<String> getStringTest1() {
return new StringTest("string test 1");
}
#bean(name="stringtest2")
public Base<String> getStringTest2() {
return new StringTest("string test 2");
}
#bean(name="integertest1")
public Base<Integer> getIntegerTest1() {
return new IntegerTest("integer test 1");
}
#bean(name="integertest2")
public Base<Integer> getIntegerTest2() {
return new IntegerTest("integer test 2");
}
I will get the expected output:
string test 1
string test 2
integer test 1
integer test 2
Any idea why my previous code only have the PostConstruct invoked for the first 2 beans?
After updating to ABP v3.0.0, I started getting DbContext disposed exception when running parallel requests to repository that create new UnitOfWork like this:
using (var uow = _unitOfWorkManager.Begin(TransactionScopeOption.RequiresNew))
I looked in the source code and found some code in AsyncLocalCurrentUnitOfWorkProvider that I don't understand. When setting current uow, it sets property in wrapper:
private static readonly AsyncLocal<LocalUowWrapper> AsyncLocalUow = new AsyncLocal<LocalUowWrapper>();
private static void SetCurrentUow(IUnitOfWork value)
{
lock (AsyncLocalUow)
{
if (value == null)
{
if (AsyncLocalUow.Value == null)
{
return;
}
if (AsyncLocalUow.Value.UnitOfWork?.Outer == null)
{
AsyncLocalUow.Value.UnitOfWork = null;
AsyncLocalUow.Value = null;
return;
}
AsyncLocalUow.Value.UnitOfWork = AsyncLocalUow.Value.UnitOfWork.Outer;
}
else
{
if (AsyncLocalUow.Value?.UnitOfWork == null)
{
if (AsyncLocalUow.Value != null)
{
AsyncLocalUow.Value.UnitOfWork = value;
}
AsyncLocalUow.Value = new LocalUowWrapper(value);
return;
}
value.Outer = AsyncLocalUow.Value.UnitOfWork;
AsyncLocalUow.Value.UnitOfWork = value;
}
}
}
private class LocalUowWrapper
{
public IUnitOfWork UnitOfWork { get; set; }
public LocalUowWrapper(IUnitOfWork unitOfWork)
{
UnitOfWork = unitOfWork;
}
}
That does not look thread-safe, as any thread can set new UnitOfWork and then dispose it.
Is it a bug? I can use my own implementation of ICurrentUnitOfWorkProvider, without wrapping, but I'm not sure if that is correct.
Update
I can't give an example with DbContext disposed exception, but here is one with null reference exception in repository.GetAll() method. I think it has the same reason.
namespace TestParallelEFRequest
{
public class Ent1 : Entity<int> { public string Name { get; set; } }
public class Ent2 : Entity<int> { public string Name { get; set; } }
public class Ent3 : Entity<int> { public string Name { get; set; } }
public class Ent4 : Entity<int> { public string Name { get; set; } }
public class DomainStartEvent : EventData {}
public class DBContext : AbpDbContext {
public DBContext(): base("Default") {}
public IDbSet<Ent1> Ent1 { get; set; }
public IDbSet<Ent2> Ent2 { get; set; }
public IDbSet<Ent3> Ent3 { get; set; }
public IDbSet<Ent4> Ent4 { get; set; }
}
public class TestService : DomainService, IEventHandler<DomainStartEvent>
{
private readonly IRepository<Ent1> _rep1;
private readonly IRepository<Ent2> _rep2;
private readonly IRepository<Ent3> _rep3;
private readonly IRepository<Ent4> _rep4;
public TestService(IRepository<Ent1> rep1, IRepository<Ent2> rep2, IRepository<Ent3> rep3, IRepository<Ent4> rep4) {
_rep1 = rep1;_rep2 = rep2;_rep3 = rep3;_rep4 = rep4;
}
Task HandleEntityes<T>(IRepository<T> rep, int i) where T : class, IEntity<int> {
return Task.Run(() => {
using (var uow = UnitOfWorkManager.Begin(TransactionScopeOption.RequiresNew)) {
Thread.Sleep(i); // Simulating work
rep.GetAll(); // <- Exception here
}
});
}
public void HandleEvent(DomainStartEvent eventData) {
using (var uow = UnitOfWorkManager.Begin(TransactionScopeOption.RequiresNew)) {
var t1 = HandleEntityes(_rep1, 10);
var t2 = HandleEntityes(_rep2, 10);
var t3 = HandleEntityes(_rep3, 10);
var t4 = HandleEntityes(_rep4, 1000);
Task.WaitAll(t1, t2, t3, t4);
}
}
}
[DependsOn(typeof(AbpEntityFrameworkModule))]
public class ProgrammModule : AbpModule
{
public override void PreInitialize() { Configuration.DefaultNameOrConnectionString = "Default"; }
public override void Initialize()
{
IocManager.RegisterAssemblyByConvention(Assembly.GetExecutingAssembly());
Database.SetInitializer<DBContext>(null);
}
}
class Program {
static void Main(string[] args) {
using (var bootstrapper = AbpBootstrapper.Create<ProgrammModule>()) {
bootstrapper.Initialize();
bootstrapper.IocManager.Resolve<IEventBus>().Trigger(new DomainStartEvent());
Console.ReadKey();
}
}
}
}
I have an MVC4 site that uses both MVC and WebAPI in it. All was going well till I tried to change my classes to have a cross cutting AOP class that would help with caching data. I am now finding that when I call a method that does not have the InterceptAttribute on it, it will crash because Ninject didn't inject with a parameter, and it fails.
My BLL class looks like this:
public class FooBLL
{
#region Private Variables
private readonly IDAL _context;
#endregion
#region Constructor
public Foo(IDAL context)
{
_context = context;
}
#endregion
#region Public Methods
public List<Bar> GetAllBars()
{
return _context.GetAllBars();
}
public List<Bar> GetTwoBars()
{
return _context.GetTwoBars();
}
#endregion
}
My WebApi Controller looks like this:
public class FooController : ApiController
{
#region Private Variables
private readonly FooBLL _fooBll;
#endregion
#region Constructor
public FooController(FooBLL fooBll)
{
_fooBll = fooBll;
}
#endregion
#region Public Methods
#region Estimate Types
#region Get
public List<Bar> GetAllBars()
{
return _fooBll.GetAllBars();
}
public List<Bar> GetTwoBars()
{
return _fooBll.GetTwoBars();
}
#endregion
#endregion
#endregion
}
In my Website, I created the following Ninject classes for resolving the controllers:
public class NinjectRegistrations : NinjectModule
{
public override void Load()
{
Kernel.Bind<IDAL>().To<DAL>().InSingletonScope();
}
}
public class NinjectDependencyResolver : NinjectDependencyScope, IDependencyResolver, System.Web.Mvc.IDependencyResolver
{
private readonly IKernel kernel;
public NinjectDependencyResolver(IKernel kernel)
: base(kernel)
{
this.kernel = kernel;
}
public IDependencyScope BeginScope()
{
return new NinjectDependencyScope(this.kernel.BeginBlock());
}
}
public class NinjectDependencyScope : IDependencyScope
{
private IResolutionRoot resolver;
internal NinjectDependencyScope(IResolutionRoot resolver)
{
Contract.Assert(resolver != null);
this.resolver = resolver;
}
public void Dispose()
{
var disposable = this.resolver as IDisposable;
if (disposable != null)
{
disposable.Dispose();
}
this.resolver = null;
}
public object GetService(Type serviceType)
{
if (this.resolver == null)
{
throw new ObjectDisposedException("this", "This scope has already been disposed");
}
return this.resolver.TryGet(serviceType);
}
public IEnumerable<object> GetServices(Type serviceType)
{
if (this.resolver == null)
{
throw new ObjectDisposedException("this", "This scope has already been disposed");
}
return this.resolver.GetAll(serviceType);
}
}
In Global.asax I then register this resolver:
NinjectHelper.Kernel = new StandardKernel(modules);
var ninjectResolver = new NinjectDependencyResolver(NinjectHelper.Kernel);
DependencyResolver.SetResolver(ninjectResolver); // MVC
GlobalConfiguration.Configuration.DependencyResolver = ninjectResolver; // Web API
//Register Filter Injector
GlobalConfiguration.Configuration.Services.Add(typeof(System.Web.Http.Filters.IFilterProvider), new NinjectWebApiFilterProvider(NinjectHelper.Kernel));
Things were fine till I added the attribute Cache using Ninject.Extensions.Interception.Attributes.InterceptAttribute.
The class now looks like this (note that I added a parameterless constructor and marked one of the methods as virtual, these are both required for the Interception to work):
public class FooBLL
{
#region Private Variables
private readonly IDAL _context;
#endregion
#region Constructor
public Foo(IDAL context)
{
_context = context;
}
public Foo()
{
}
#endregion
#region Public Methods
public List<Bar> GetAllBars()
{
return _context.GetAllBars();
}
[Cache(DefaultTimeoutMinutes = 20)]
public virtual List<Bar> GetTwoBars()
{
return _context.GetTwoBars();
}
#endregion
}
Now on the WebAPI controller, when I call GetToBars(the method with the Intercept Attribute), everything still works fine.
However, when I call GetAllBars(the method that doesn't have the Intercept Attribute), I fail with an exception that _context is null.
Any help would be greatly appreciated.
Ben
I am using Umbraco 6.1 with an UmbracoApiController which has a IUnitOfWork injected into it's constructor. To inject the dependencies, I am using Unity, like I have in the past with standard Web API projects. Normally, I set unity up in the Global.asax.cs. As Umbraco does not have this I have created my own UmbracoEvents handler, which inherits from IApplicationEventHandler, and has the methods:
OnApplicationInitialized
OnApplicationStarting
OnApplicationStarted
ConfigureApi
In the OnApplicationStarted method I set up my EF database, db initializer etc and call ConfigureApi to set up Unity. My OnApplication Started and ConfigureApi methods looks like this:
public void OnApplicationStarted(UmbracoApplicationBase umbracoApplication, ApplicationContext applicationContext)
{
_applicationContext = applicationContext;
_umbracoApplication = umbracoApplication;
_contentService = ApplicationContext.Current.Services.ContentService;
this.ConfigureApi(GlobalConfiguration.Configuration);
Database.SetInitializer(null);
PropertySearchContext db = new PropertySearchContext();
db.Database.Initialize(true);
}
private void ConfigureApi(HttpConfiguration config)
{
var unity = new UnityContainer();
unity.RegisterType<PropertiesApiController>();
unity.RegisterType<IUnitOfWork, UnitOfWork>(new HierarchicalLifetimeManager());
config.DependencyResolver = new IoCContainer(unity);
}
My Controller code:
public class PropertiesApiController : UmbracoApiController
{
private readonly IUnitOfWork _unitOfWork;
public PropertiesApiController(IUnitOfWork unitOfWork)
{
if(null == unitOfWork)
throw new ArgumentNullException();
_unitOfWork = unitOfWork;
}
public IEnumerable GetAllProperties()
{
return new[] {"Table", "Chair", "Desk", "Computer", "Beer fridge"};
}
}
My Scope Container/IoC Container code: (as per http://www.asp.net/web-api/overview/extensibility/using-the-web-api-dependency-resolver)
public class ScopeContainer : IDependencyScope
{
protected IUnityContainer container;
public ScopeContainer(IUnityContainer container)
{
if (container == null)
{
throw new ArgumentNullException("container");
}
this.container = container;
}
public object GetService(Type serviceType)
{
if (container.IsRegistered(serviceType))
{
return container.Resolve(serviceType);
}
else
{
return null;
}
}
public IEnumerable<object> GetServices(Type serviceType)
{
if (container.IsRegistered(serviceType))
{
return container.ResolveAll(serviceType);
}
else
{
return new List<object>();
}
}
public void Dispose()
{
container.Dispose();
}
}
public class IoCContainer : ScopeContainer, IDependencyResolver
{
public IoCContainer(IUnityContainer container)
: base(container)
{
}
public IDependencyScope BeginScope()
{
var child = this.container.CreateChildContainer();
return new ScopeContainer(child);
}
}
My IUnitOfWork code:
public interface IUnitOfWork : IDisposable
{
GenericRepository<Office> OfficeRepository { get; }
GenericRepository<Property> PropertyRepository { get; }
void Save();
void Dispose(bool disposing);
void Dispose();
}
My UnitOfWork implementation:
public class UnitOfWork : IUnitOfWork
{
private readonly PropertySearchContext _context = new PropertySearchContext();
private GenericRepository<Office> _officeRepository;
private GenericRepository<Property> _propertyRepository;
public GenericRepository<Office> OfficeRepository
{
get
{
if (this._officeRepository == null)
{
this._officeRepository = new GenericRepository<Office>(_context);
}
return _officeRepository;
}
}
public GenericRepository<Property> PropertyRepository
{
get
{
if (this._propertyRepository == null)
{
this._propertyRepository = new GenericRepository<Property>(_context);
}
return _propertyRepository;
}
}
public void Save()
{
_context.SaveChanges();
}
private bool disposed = false;
public virtual void Dispose(bool disposing)
{
if (!this.disposed)
{
if (disposing)
{
_context.Dispose();
}
}
this.disposed = true;
}
public void Dispose()
{
Dispose(true);
GC.SuppressFinalize(this);
}
}
I have used unity/DI with MVC4/WebAPI controllers and this implementation of UnitOfWork many times before without issue, so I'm thinking it's Umbraco specific.
I have also debugged the application and made sure that it hits OnApplicationStarted and that its parameters are not null.
The GetAllProperties method in the controller is just a test method to make sure it is all working fine, however, when I try and access this action I get the error:
"The type IUnitOfWork does not have an accessible constructor"
Does anyone have experience with using Umbraco 6.1 and it's UmbracoApiController with dependency injection/Unity?
Also, on an unrelated note, is there a way to return JSON instead of XML in the action? In Web API you would just define the formatter in the WebApi.config but there is none in Umbraco.
Thanks,
Justin
In case you haven't found a solution to your problem? Download this nuget package and right after building your unity container:
GlobalConfiguration.Configuration.DependencyResolver =
new Unity.WebApi.UnityDependencyResolver(Bootstrapper.Container);
Notice the namespace which is different than Unity.Mvc4.UnityDependencyResolver.