Archives for June, 2009
I don’t know whether I just noticed this too late, but the SaveChanges method in upcoming EF4 is virtual. This is a great small change.
You can generate (if you’re using T4 templates for generating) your own SaveChanges method and do some work there. Well, before actions can be done using SavingChanges event, i.e. validation like in this example. But for after actions, there’s no SavedChanges event (neither in EF4). Hence adding some code into this method is very handy. For example logging actions performed is super easy.
This small change opens up new ways of thinking about SaveChanges usage.
The method also have new SaveOptions parameter. As it’s enum we can expect adding more options in future versions.
The Beta 1 of Entity Framework v4 (confused with versions?) is out for a while and you may be tempted to check the new features comming. And why not with Firebird.
Good news is, that there’s no breaking change right now I’m aware of. Hence you can easily jump into VS2010, install DDEX and test it. Of course, some new features are now not fully functioning as it’s beta and/or because it’s not nailed down for provider writers (i.e. DDL script generation). But as far as something will be ready, I’ll try my best to get it into Firebird’s provider.
On the other hand, even the new singularize/pluralize function is nice and works with all databases, you still get all names in uppercase, only the added suffixes are lowercase. And that looks even more weird. You can fix it yourself manually or let computer do the work, but still wondering why there isn’t any option to fix this (as uppercased name is default when not quoted according to SQL standard).
Anyway, if you found any problems with EF4, let me know.
Pregenerating views can speed up startup time of you application. You may find more info in my previous post about it. On the other hand, setting up this for all developers or i.e. build machine can be tricky. Thus it may be worth to think whether it’s for you or not.
One reason not to pregenerate these views (besides laziness 
) could be, that your application is long running (i.e. windows service or ASP.NET app (I’m not IIS and ASP.NET guru so I hope this is also the case)) and if these views would be cached in smart way, application will be benefit from it even without pregenerating.
Good news. These bi-directional views are cached per appdomain. So if you don’t care about startup time and your application is long running (in one appdomain) you don’t have to care about setting up all the stuff described in the article.
Vista mi jaksi vypíná grafika-3D-cosi (zajímavé, že to začalo zlobit až po instalaci systému s integrovaným SP1). Takže se minimalizace okna, scrollování nebo třeba Alt-Tab trošku zasekávalo. Hlavně Alt-Tab mě štvalo, protože trvalo relativně dlouho, než se zobrazilo. Po dlouhé snaze vyřešit to systémově, jsem přešel na “programátorské” řešení.
Vzal jsem první WPF sample s pěknou 3D animací a udělal z toho aplikaci, bez tlačítka na taskbaru. A překvapivě to skvěle funguje. Aplikace mi běží na pozadí a jsem spokojenej.
Zde je vidět desktop s běžící aplikací:
Few days ago I got interesing information that IBM in AppScan Standard Edition is using Firebird database and FirebirdClient as .NET provider. It’s a great finding where all is Firebird used.
Do you have any interesting usage of Firebird and/or FirebirdClient? Let us (or me) know.
A lot of people get used to understand the ObjectStateManager class. Yep, it’s the place where all the change tracking magic happens. But there’s also one other place that’s interesting. It’s MetadataWorkspace. With this class’ properties and methods you can find almost everything about stuff in your model, in runtime.
The information inside MetadataWorkspace is among others organizes into so-called spaces (DataSpace enum). Probably CSpace and SSpace are the most used. The CSpace contains stuff you have in your conceptual model (what you see in designer) and SSpace contains stuff from store model, so the database “structure”.
The items in MetadataWorkspace can be anything from the Metadata Type Hierarchy (definitely look at this link) starting with GlobalItem. This contains i.e. EntityType, ComplexType or EdmFunction. These types inside are containing a lot of other useful information. Thus when you want to show i.e. all tables and primary keys for these tables you can use (don’t forgot to first, for example, query for one record to load metadata or use this trick):
MetadataWorkspace mw = ent.MetadataWorkspace;
var items = mw.GetItems<EntityType>(DataSpace.SSpace);
foreach (var i in items)
{
Console.WriteLine("Table Name: {0}", i.Name);
Console.WriteLine("Keys:");
foreach (var key in i.KeyMembers)
{
Console.WriteLine("t {0} ({1})", key.Name, key.TypeUsage.EdmType.FullName);
}
}
I’m first asking for all entity types from store space, which gives us all tables there. Then I’m simply inspecting the properties to get some useful info – first the name of table and then all key members reading name and type (EdmType). Also the MetadataProperties is definitely worth looking (see below). It’s just a tip of an iceberg, and that was not difficult. You can literally get any information you’ve ever seen in your model.
And not only what you’ve seen in your model directly. For instance, the provider may expose some store specific functions you can use in ESQL queries. So the code:
var functions = mw.GetItems<EdmFunction>(DataSpace.SSpace);
foreach (EdmFunction function in functions.Where(f => (bool)f.MetadataProperties["BuiltInAttribute"].Value))
{
Console.WriteLine(function.FullName);
}
Gives you all of these functions. If you remove the condition you’ll get all functions, including stored procedures of user defined functions. Wanna parameters of these functions? No problem, it’s there. Just check Parameters property.
And that’s only the SSpace. The conceptual part of model is described there too. To see how much is the model interconnected (i.e. to consider pregenerating views) simply run:
var entities = mw.GetItems<EntityType>(DataSpace.CSpace);
foreach (var entity in entities)
{
Console.WriteLine("Table Name: {0}", entity.Name);
Console.WriteLine("Navigation Properties#: {0}", entity.NavigationProperties.Count);
}
Interesting may be also the object model aka objects created from conceptual model. Although I can continue with some more or less useful examples – to see all to posibilities (and also hit the accessibility limitations), try it yourself. And if you stuck, feel free to ask (here or in forums).
Jarek Kowalski posted to MSDN Code Gallery wrappers for any ADO.NET Entity Framework provider with ability to do caching and logging.
For me especially the logging (EFTracingProvider) is interesting, as I’m always checking queries sent to database. For selects it’s plausible with ToTraceString method (but not with i.e. query.First()), but looking for CUD is not so easy. With EF provider for Firebird, you can check, by default, Output window of Visual Studio, where all queries are placed or use any listener to write these i.e. to the file (this logging is only in debug builds of FirebirdClient).
I was thinking about creating some wrapped provider, but this is better (and without work ).
When you write huge (I mean really huge, not a ~300 lines baby) complex query you may experience, that the generated query looks (well) suboptimal. The reason is that after some line the Entity Framework will gave up on optimization and simply throws it as is. To turn this off and force to simplify it whatever query it is you can add this into your app.config, as Kati Iceva pointed in MSDN forums.
<system.diagnostics>
<switches>
<add name="System.Data.EntityClient.IgnoreOptimizationLimit" value="1" />
</switches>
</system.diagnostics>
But also take into account that this may end up with a significant time spent in EF’s code and at the end the query will be still too big for the store to run it or the store would do better job optimize it. Thus the overall time may be same or even worse (= use carefully).
It’s also worth, if you’re able to write the query a lot simpler, to discuss this in MSDN forums, so the code in EF can be improved.
Nejsem žádný purista co se týká jazyka. Sám jej slušně przním anglickými výrazy. Co mě ale v poslední době hodně vytáčí je neschopnost některých napsat správně některá základní slova. A v posledním půl roce jsem toto viděl více než často.
Co je tak těžkého na rozlišení mně a mě? Hlavně správné použití mně. Když napíšu “Mě se to líbí.”, je to špatně, špatně, špatně. Má být “Mně se to líbí.”.
Když něco píšete, snažte se aspoň trochu nad výsledkem zamyslet. Tak a doufám, že tu nemám hrubku.
Yesterday I wrote about new CountdownEvent class. But what’s better than see some example of usage?
Below is pretty simple example of usage. You can see, it’s very similar to work with array of i.e. ManualResetEvent. But you have also some handy methods and properties. For instance: AddCount/TryAddCount or CurrentCount. Very handy.
class Program
{
static void Main(string[] args)
{
using (CountdownEvent cde = new CountdownEvent(10))
{
for (int i = 0; i < cde.InitialCount; i++)
{
new Thread(new ParameterizedThreadStart(Dummy)).Start(cde);
//ThreadPool.QueueUserWorkItem(new WaitCallback(Dummy), cde);
}
cde.Wait(2000);
Console.WriteLine("Threads done in first 2 seconds: {0}.", cde.InitialCount - cde.CurrentCount);
cde.Wait();
Console.WriteLine("All threads done.");
}
}
static void Dummy(object o)
{
Thread.Sleep(new Random().Next(5000));
(o as CountdownEvent).Signal();
}
}
As I said, the work is similar to work with array of ManualResetEvent, just packed into nicer cake. In fact, if you start ILDasm and look into the code you’ll see, that’s implemented very similarly. It’s using ManualResetEventSlim (also new in .NET 4) internally to signal and smart work with Interlocked class do decrement (or increment) the number of signals received.
Do you like the class too?
Today, while just randomly walking thru MSDN documentation I found new CountdownEvent class. It’s nothing ultra special. You can write similar class yourself in a couple of hours. But it’s great that you don’t have to. And with all the new stuff, like Task object, writing the multi-threaded apps is more easier than before (but sometimes it’s neat to just write some algorithm/problem using only thread and critical sections).
Having girls/womans in a dev team is always fun. Yes, you have to wash your body more than every Visual Studio release. 
But girls have also different way of looking at the problem. I had this experience in a training last week. There’s one girl in dev team. While showing how to work with Cryptography namespace I was creating simple method for SHA1 hash. Saying, that’s pretty easy to create SHA1 hash in a couple of lines I got a question (yes, from the girl ), whether it’s possible to do it in one row, to make code more compact and show that’s double pretty easy. And it really is:
label1.Text = string.Join(string.Empty, new SHA1Managed().ComputeHash(Encoding.Unicode.GetBytes(textBox1.Text)).Select(_ => _.ToString()).ToArray());
It’s not the most readable code, but as a helper function somewhere …
Minulý týden jsem byl v Liberci, kde jsem prováděl školení pro firmu INISOFT. Mimochodem skvělá parta lidí – a BabySmash, náhodou zmíněný, zaznamenal fenomenální úspěch.
Mimo to jsem měl možnost se podívat na horu Ještěd (1012m). Na vrcholu je stavba (televizní vysílač a hotel), jejíž tvar má být rotační hyperboloid. Z dálky možná, ale na místě … Některé plochy jsou prostě víc rovné než “oblé”. Podvod. Nicméně i tak doporučuji navštívít. Výhled je super a stavba opravdu pěkně následuje tvar terénu.
