Monday, December 27, 2010

State and CycledEnumerator

The GoF State pattern is not on top of my list of favorites (Decorator is), but I am using it from time to time. And as you probably are aware there is this question about the implementation of State: whoever will switch between states and when?

I thought that Iterator -- another GoF pattern -- might be a good example of something that encapsulates state switching logic in some cases, when the states are switched strictly in the order you know beforehand. I realize that not every State implementation is that simple, but if yours is, you might want to use the idea: just pass the IEnumerable of IState to the implementation of State, and use MoveNext() and Current {get;} on the Enumerator internally to switch states. If you are doing something like this, it might be a good idea to jump back to State #1 after you've reached State #Last. I've written a little helper class to do just that:

using System;
using System.Collections.Generic;

namespace Primitives
{
    public class CycledEnumerator<T> : IEnumerator<T>
        where T : class
    {
        private LinkedList<T> _chainedObjects;
        private LinkedListNode<T> _currentObject;

        public CycledEnumerator(IEnumerable<T> objectsToChain)
        {
            if (objectsToChain == null)
            {
                throw new ArgumentNullException("objectsToChain");
            }

            _chainedObjects = new LinkedList<T>(objectsToChain);

            if (_chainedObjects.Contains(null))
            {
                throw new ArgumentException("Nulls are not allowed.", "objectsToChain");
            }

            if (_chainedObjects.Count < 1)
            {
                throw new ArgumentException("Need at least one item in the set.", "objectsToChain");
            }

            this.MoveBeforeFirstElement();
        }

        private void MoveBeforeFirstElement()
        {
            _currentObject = null; // I don't like nulls at all, but I'm putting it here for the sake of consistency with Microsoft's implementations of Enumerator (which is a variant of GoF Iterator pattern).
        }

        public T Current
        {
            get
            {
                return _currentObject.Value;
            }
        }

        public void Dispose()
        {
        }

        object System.Collections.IEnumerator.Current
        {
            get
            {
                return _currentObject;
            }
        }

        public bool MoveNext()
        {
            if ((_currentObject == null) || (_currentObject.Next == null))
            {
                MoveToFirstElement();
            }
            else
            {
                _currentObject = _currentObject.Next;
            }

            return true;
        }

        private void MoveToFirstElement()
        {
            _currentObject = _chainedObjects.First;
        }

        public void Reset()
        {
            this.MoveBeforeFirstElement();
        }
    }
}

(That said, I don't like Iterator much (and all of its buddies, like the Enumerator): it exposes too much of its state to the client, and has too complex a contract. It's all like "if MoveNext() returns true, then Current contains the next element", or: "after initialization the Enumerator is positioned before the first element, and you should call MoveNext() to get anything from it".)

Gang Of Four Patterns

Some days I find myself really believing that Gang Of Four patterns describe about 90% of all domain-model-related software design tasks out there. On our last design meeting I only had to name the applicable patterns, and that about did it: "Huh, this can be done with a Visitor", or "Sounds like an ordinary Composite to me"... That is not to say I'm good, but rather to reiterate what this guy told me once: "Everything has already been written".

("Design Patterns: Elements of Reusable Object-Oriented Software" was one of the first books I've read on the subject, years ago, and it's a really good read.)

Monday, August 23, 2010

Databases vs. everchanging model

This sprint I've been struggling hard to postpone the creation of a database to back up WCF service operations 'till next sprint. Last sprint I've insisted on keeping the store of service objects in memory - so the service loses all data upon restart. This was enough to make an iteration (and I think it was a good decision) but a lot of people on the team pushed back.

The reason I don't want a database from day 1 is our domain model is undergoing such changes that it would have taken us an entire sprint redesigning the database for every change, if we had decided to go down that road. We've spent that time on useful things instead, I hope.

Last week we, again, came to a point where a decision was needed on whether to create the database or not. And once again I pushed back, so we've chosen to save all data into a single file (having more than one would require extra work for reconstructing object references) using plain old data contracts. A question came up though: how do you save all those different collections of domain model elements preserving object references between those collections? The answer to that was dumping all objects into a List<object>, creating a DataContractSerializer around it (with exactly that type - List<object>), feeding to the serializer a list of known types made of all types of contracts (note that it's no longer domain model elements) that are used to store the state of the service, telling the serializer to preserve object references. Of course, all of this would still be hard to do (because of reference fixups), if we had not decided to use data contracts everywhere for the state of the service - as opposed to domain model objects we had in sprint 1.

Note that this is a temporary solution, we will migrate to a database at some point. That said, I suppose modifying data contracts is slightly easier than modifying the database -- I may be exaggerating the complexity of making changes to the databases since I haven't seen any database refactoring tools, and there must be some.

(When putting the reconstructed object graph' elements back to their collections we've iterated thru deserialized elements and added them to appropriate collections depending on the type of element).

Could post the code proving the concept works, if anyone is interested.

Saturday, June 26, 2010

Generics

Generics this generics that. Surely, generics were one of the main hits of .Net FW 2.0 and you all know the advantages they give you, but what about disadvantages? The books I've read on software development are hardly giving anything for disadvantages, and that's suspicios because nothing is ideal and if you gain something, that means losing something the very same moment you gain.

The pain in the ass generics give you can be described as follows:
- generic classes and interfaces with more than two type parameters are hard to understand and maintain;
- generic class, say, List<B> is not convertible to List<A> if B is subclass of A (as of .Net FW 3.5), moreover there is no direct relationship between List<A> and List<B> which could be described in terms of OO paradigm;
- generics cypher the code to some extent: when you see an instance of Mine.MyGenericType<Mine.MyType1, Mine.MyType2, Mine.MyType2> in code you have to scratch your head to figure out which parameter means what, because they only have names internal to MyGenericType and don't mean anything outside of that class (this is somewhat less of a problem when you are writing the code, thanks to intellisense);
- generics used solely for type safety scare me a bit: the main drivers for introducing generics were code reuse, type safety and performance (in that order). There is usually a way or two to avoid using generics without losing type safety.

That's my humble opinion and don't get me wrong, I absolutely love generics. Just make sure you don't rely on them too much.

Wednesday, May 26, 2010

Buggy trace listener

Recently a colleague of mine ran into the troublesome TextWriterTraceListener. As the name suggests, it's a trace listener that writes trace messages to a file. The way you use it, as any other listener, is as follows: either you instantiate the listener in code and add it to the static Trace.Listeners collection, or let the runtime do the same thing for you by saying so in App.config -- then you dispatch trace messages to this and other trace listeners hooked up, thru the static Trace class.

The TextWriterTraceListener and the Trace class work together just fine, as long as you don't try to use them from a finalizer when your application is terminating. You say "Trace.Writeline("Hi John, got some bad news for you!")" in the finalizer of your class which happens to run -- among other cases -- during application shutdown, and doh! -- you get a marvelous error saying that the FILE has already been closed. Isn't it just nice?

(The listener references a FileStream which has a finalizer of its own and finalizers run in no particular order: chances are your finalizer (calling into Trace) will be out of luck, calling into a dead trace listener. There are several ways in which you can make that listener do a better job, but I ended up not using Trace and TextWriterTraceListener at all. This is definitely a design flaw in both Trace and that particular listener. It should be fixed from the inside, I don't want to work around this).

Friday, April 9, 2010

AppDomains and true isolation

As you may very well know, .Net Application domains (or AppDomains for short) are the "lightweight processes", according to authors of several books on .Net Framework (FW) I've read, and -- to a great many developers I've spoken to. For most part I agree with the metaphor, but there is one thing the AppDomains do not provide you with, while processes do: fault tolerance. The operating system doesn't crash if one of its processes crashes. But as of .Net FW 2.0 an unhandled exception in any AppDomain in a process brings that process down. I tend to think in terms of corner cases, worst case scenarios (my analyst complains about that a lot), so for me this basically means that Appdomains hardly provide any isolation at all.

Let's look at the ways of fixing that. One way to go would be to create a custom CLR host which would override the standard policy for unhandled exceptions. I will not cover this option as I'm not an expert at customizing CLR hosts. Another way is to revert to .Net FW 1.0/1.1 policy for unhandled exceptions, AND (this is very important) to emulate .Net FW 2.0 policy for the main Appdomain of the application, and to unload other domains, should they encounter an unhandled exception. This is the way I'm going to show and explain.

First, let's prove that the problem exists. The following code crashes the process for good (see for yourself by compiling it in a console application). Run it without the debugger attached -- to not be stuck on the line where the exception is being thrown:
class Program
{
    private static AppDomain _subDomain;

    static void Main(string[] args)
    {
        _subDomain = AppDomain.CreateDomain("CodeRunningDomain");

        try
        {
            _subDomain.DoCallBack(delegate()
            {
                new Thread(delegate()
                {
                    throw new InvalidOperationException();
                }).Start();
            });
        }
        catch
        {
            // this block is for demonstration purposes only,
            // generally it's not a good idea to catch System.Exception.
            // this block is never hit, of course
        }

        Thread.Sleep(TimeSpan.FromDays(2)); // the process is torn down faster than in 2 days, you will see.
    }
}

Not much to explain here -- an unhandled exception by now-standard .Net policy tears down the process. I agree that this is a great default policy due to possible state corruptions related to an error that went unhandled. But in case you're building a host, say for running plugins in their own AppDomains, or trying to execute a piece of code you don't trust in its own AppDomain, you don't want your host to die because a plugin has errored out, or because the code you did not trust, proved right your assumptions about it.

First step to work around the issue is to put the crappy .Net FW 1.0/1.1 unhandled exception policy back to business. According to this policy, the process is not torn down, when an exception goes unhandled, as you may know. To achieve this, paste the following to a newly created application configuration file for the Console Application you've probably created to run the code above:
<?xml version="1.0" encoding="utf-8" ?>
<configuration>
  <runtime>
    <legacyUnhandledExceptionPolicy enabled="true"/>
  </runtime>
</configuration>

Now everything runs without crashing and the program will block for years, if you let it. But that's not what we want: .Net FW 2 policy for exceptions happens to be cool, we want it back, but only for the main AppDomain. Here is what we do: we hook up the handlers for every unhandled exception in every AppDomain. When an exception propagates up the main AppDomain (it does propagate), we look which domain has it originated from, if it's not from the main AppDomain, we unload the offending AppDomain and carry on. Otherwise we tear the process down, as (not exactly, but close) the CLR does in these cases. Of course, we're working under the assumption that a failure in one AppDomain does not (and mustn't) affect the other domains. Keep in mind that this bold assumption is not always true. I don't want to bore you to death, here is the code. (Again: run it without the debugger attached -- to not be stuck on the line where the exception is being thrown.)

using System;
using System.IO;
using System.Threading;

namespace ProperAppDomainIsolation
{
    class Program
    {
        private static AppDomain _subDomain;

        static void Main(string[] args)
        {
            AppDomain.CurrentDomain.UnhandledException +=
              new UnhandledExceptionEventHandler(MainDomain_UnhandledException);

            _subDomain = AppDomain.CreateDomain("CodeRunningDomain");

            _subDomain.UnhandledException +=
              new UnhandledExceptionEventHandler(Subdomain_UnhandledException);

            try
            {
                _subDomain.DoCallBack(delegate()
                {
                    new Thread(delegate()
                        {
                            throw new InvalidOperationException();
                        }).Start();
                });
            }
            catch (Exception)
            {
                // log, throw, do something.
            }

            Thread.Sleep(TimeSpan.FromDays(2));
        }

        static void MainDomain_UnhandledException(object sender, UnhandledExceptionEventArgs e)
        {
   // runs in main AppDomain
            Exception exception = e.ExceptionObject as Exception;

            if (exception == null)
            {
                Environment.FailFast("Very descriptive message");
            }

            Console.WriteLine("Got an exception in main application domain.");

            try
            {
                Marker exceptionMarker = ExceptionMarker.GetExceptionMarker(exception);

                Console.WriteLine("Retrieved marker: " + exceptionMarker.Value);
                Console.WriteLine("Offending AppDomain: " + _subDomain.Id);

                if (exceptionMarker.Value.Equals(_subDomain.Id.ToString(), StringComparison.Ordinal))
                {
                    new Thread(delegate()
                        {
                            try
                            {
                                Console.WriteLine("Unloading the offending application domain...");

                                AppDomain.Unload(_subDomain);
                                _subDomain = null;

                                Console.WriteLine("Unloaded the offending application domain...");
                            }
                            catch (Exception ex)
                            {
                                Console.WriteLine(ex.ToString());
                            }
                        }).Start();
                }
            }
            catch (InvalidOperationException)
            {
                // exception is not marked - originates from the main application domain. Demonstration code - a more specific exception type is required.
                Environment.FailFast("Terminating to an unhandled exception in the main Application Domain.");
            }
        }

        static void Subdomain_UnhandledException(object sender, UnhandledExceptionEventArgs e)
        {
   // runs in subdomain
            Exception exception = e.ExceptionObject as Exception;

            if (exception == null)
            {
                Environment.FailFast("Very descriptive message");
            }

            ExceptionMarker.MarkException(exception, new Marker(AppDomain.CurrentDomain.Id.ToString()));
        }
    }
}

On my box the output of the program is as follows:

Got an exception in main application domain.
Retrieved marker: 2
Offending AppDomain: 2
Unloading the offending application domain...
Unloaded the offending application domain...

Now we've got ourselves a host! The only fun part here is in distinguishing between the exceptions that came from the main AppDomain and all other domains.

(ExceptionMarker adds a special marker to the Exception's property bag - in the subdomain, and then reads the marker in main domain; if there is no marker attached to exception, then it originates from the main AppDomain, and the process is doomed. Full source code is here.)

P.S. Originally I thought that attaching a configuration file (like the one shown above) to the domains where the code runs, would suffice. Plus I planned on taking care to unload the domains when an exception occurs inside each of them. That didn't work: legacyUnhandledExceptionPolicy seems to affect the main application's AppDomain only.

Monday, February 8, 2010

lamron

"Develop a normal C# class that implements the interface you just defined".

(The fun word here is "normal".)

In overall, it's a great guideline.

Tuesday, January 5, 2010

Could not have said it better

“No one scheme of classification, more than any other, represents the real structure or order of nature. Nature indifferently submits to any and all divisions which we wish to make among existing things. Some classifications may [seem to] be more significant than others, but only by reference to our interests, not because they represent reality more accurately or adequately”.

This quote is from "Object Oriented Analysis and Design". I like the book a lot (although, it could have been shorter), and it's the best part there. If you stop to think about what it says, you will probably find that it's very true. Not only does this apply to object-oriented classification, but for any other classification or schema out there. I could talk for hours about this, but since I promised to keep everything short, I will just say: the life itself is incomprehensible.