|
0x01AA wrote: Do we really Need to write code where we Need first to solve a puzzle to get what the code is doing
Well, this is different. It's a stupid bug on my part. But it's like language. If you can speak at more than a 3rd grade level, you can express your thoughts better.
|
|
|
|
|
Quote: If you can speak at more than a 3rd grade level, you can express your thoughts better No idea what this means, sounds only I'm undeveloped
Btw. I'm aware, I'm most probably not able to solve the puzzle
It does not solve my Problem, but it answers my question
modified 19-Jan-21 21:04pm.
|
|
|
|
|
You probably could, with a bit of thinking about it - it's obvious when you see it, but it's a stinker to spot if you didn't write the code (and probably even harder if you did if you are anything like me: I tend to see what I meant to write, rather than what I did )
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
I also see what I meant to write as opposed to what I actually wrote. I've been tired and looked at two spellings of same word and get compiler errors and don't spot the error for some time.
Everyone has a photographic memory; some just don't have film. Steven Wright
|
|
|
|
|
Annoying, isn't it?
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
Yes very frustrating.
As I get older it seems to be more prevalent.
Everyone has a photographic memory; some just don't have film. Steven Wright
|
|
|
|
|
OriginalGriff wrote: I tend to see what I meant to write, rather than what I did
Yup that's me!
As an answer: I haven't gone further in the thread, but at first glance the FALSE comes from comparing 2 different instances, where the TRUE result is comparing the same instance to itself?
Fix: the first comparison is not necessary
|
|
|
|
|
Roughly 90 seconds, mostly working out which Console.WriteLine call was which.
Spoilers ahead - select the block to view:
Take a copy of the static field;
Create a new instance, thus overwriting the static field;
Compare the new instance to the copy of the old value of the static field - result = false;
Compare the new instance to the current value of the static field - result = true;
Damnit Chris, we need a <div class="spoiler"> !
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Richard Deeming wrote: select the block to view:
Now THAT is snazzy!
|
|
|
|
|
Is it a problem? Does it need to be fixed?
I wouldn't presume to say that you didn't write exactly what you intended.
Setting a static on each instantiation is unusual though.
|
|
|
|
|
The problem is that you're assigning a static variable inside the instance constructor to that instance . So every time you create an object you're changing the Context property. The reason the tests are different is because you pass the Context as a parameter so even after the update the parameter still has the old Context .
A fix would be to not do that because it's bad design I'd make Context an instance property and implement your own ==, !=, Equals, etc unless you want referential equality.
|
|
|
|
|
Just read the blog post, and I don't think your fix will work.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
It will work in that specific example because he also changed the CreateInstance call to not use conn anymore. But yes, if the parameterized constructor is called it'll still have the same issue
|
|
|
|
|
Well, it will "work", in the sense that it won't overwrite the static instance.
But it won't work, in the sense that the returned instance will be using the default connection string for the context, not the connection string from the instance passed in.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Richard Deeming wrote: But it won't work, in the sense that the returned instance will be using the default connection string for the context, not the connection string from the instance passed in.
Hmm. But the parameterless constructor uses the connection string of the (hopefully) already initialized context:
public ModelDataContext() : base(Context.Connection)
However, it's a moot point anyways as I refactored the whole mess into something much less weird.
It'll be an interesting question though to pose to a couple of the devs at work, though I'll reduce the complexity of it omitting the DataContext base class - it's rather superfluous to the example.
|
|
|
|
|
OK, so it's not quite as bad as I though.
You could still break it though:
var context1 = new ModelDataContext(new SqlConnection("server=A;database=B"));
var context2 = new ModelDataContext(new SqlConnection("server=C;database=D"));
CreateNewContext(context1, out var newConn, out var newContext);
Console.WriteLine(newConn.ConnectionString == context1.Connection.ConnectionString);
Console.WriteLine(newContext.Connection.ConnectionString == context1.Connection.ConnectionString);
Console.WriteLine(newContext.Connection.ConnectionString == newConn.ConnectionString);
Console.WriteLine(newContext.Connection == newConn);
Console.WriteLine(newContext.Connection.ConnectionString == context2.Connection.ConnectionString);
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Richard Deeming wrote: You could still break it though:
Yeah, but it's never used that way - two different connection strings. But ideally, it shouldn't be possible to use it that way.
Sigh. The irony of this is that it's code I wrote a while back that my DataContext extension methods rely on, and looking at this now, it's some serious code smell. Fortunately, fixing it affects only a couple web servers that are in operation, but I feel embarrassed. But I blame .NET's DataContext. It does way to much with regards to the state of the Table object. I get what they're trying to do, but there must be a better way that doesn't end up throwing exceptions like "this object was created in a different data context."
|
|
|
|
|
The reason for the results are pretty obvious, when a new instance is created in "CreateNewContext", the parameter "context" is still referring to the original context, not the new one ==> false; When the main program does the compare afterwards, the comparison is done to the static that holds the new context ==> true.
Fixing this is not obvious because you didn't specify what the correct behavior should be.
Do you wish it to be "True" always, or "False" always
Simply update the context parameter in the method to get True always, but I would personally recommend to re-write this whole thing, it is full of code smells. Perhaps some practice in TDD would also be helpful.
Cheers,
Anthony
|
|
|
|
|
It took about a minute to solve.
Assigning over the static for each constructor call is a WTF (as others have called out).
Way too many statics in this Program, if it was rewritten to eliminate all of the statics, then the Program instance would hold a single context and the Main would look like this:
Main(...)
{
new Program().Run();
}
If new instances of the "context" are really needed, then create a Copy constructor to reuse the default context settings on a new instance.
Or use your favorite IOC container.
|
|
|
|
|
It's the assignment of the static member from the instance constructor. HUGE no-no. If you want to make a singleton without using a DI framework, at least make the constructor private and have the static "Context" member be a property that will create a new object if one doesn't exist, and set it to a backing field. Something like this (might need to make Current a method instead of property if you need to pass constructor arguments, but you get the idea):
public class Singleton
{
private Singleton _current;
public Singleton Current
{
get
{
if (_current == null)
{
_current = new Singleton();
}
return _current;
}
}
private Singleton()
{
}
}
|
|
|
|
|
The first 'true' you receive you explicitly cast the object on the function call to a DataContext type.
The second you're comparing a ModelDataContext.Context object - which I'm not sure if it's a DataContext type - if not - then you're comparing unlike objects and the call will fail.
Try explicitly casting the ModelDataContext.Context, and for overkill the newdc to a DataContext type and do the comparison that way.
ie:
Console.WriteLine((DataContext) ModelDataContext.Context == (DataContext) newdc);
|
|
|
|
|
It is a simple newb mistake. Basically, you have written overly complex code making it hard for you to debug. Simplify the syntax as much as possible and the symantec issues will reveal themselves
|
|
|
|
|
...enough to personalise each email just for me and our cat Dij:
Quote: [~forename~}, Treat Your Kitty To The Al La Carte Menu, Minus The Hefty Bill
...
Dear Paul,
...
Typing lessons may help, chaps and chapesses ...
Bad command or file name. Bad, bad command! Sit! Stay! Staaaay...
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
well at least their description of Bill is PC.
This internet thing is amazing! Letting people use it: worst idea ever!
|
|
|
|
|
That's a nice one, [~forename~}!
Bastard Programmer from Hell
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|