|
Are you creating the expression tree every time you need a new instance of one of the classes, or are you caching the compiled Func<object[], object> for each class type?
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
I'm caching the Func<object[], object>'s in a dictionary. Just seems slower then doing a normal new up for the same # of iterations and I was under the impression its supposed to be just about the same performance. I realize my code has other overhead, of course, but still...
|
|
|
|
|
Seems like it might have something to do with the parameters. Right now the constructor for the test class is:
public Test(string str, int i, Color color)
{
}
and new'ing up 1M takes 475ms with my current code. By doing NOTHING but changing the constructor to just:
public Test(/*string str, int i, Color color*/)
{
}
it drops to 200ms. So I tried adding 1 param back and it jumps to 312ms. 2 params = 390ms.
So... every additional param is adding quite a bit of time  .
Now, adding back the first param was +112ms, but the 2nd one was only an additional +78ms. I would expect adding back the first param would be expensive since it would now travel down the param code path where without params it skips all that. However, I expected adding the 2nd param would be much cheaper since its just adding another iteration to the param loop.
A standard new up with all 3 params for 1M iterations takes only 31ms.
I'm estimating that there is about 140ms of overhead built into my dynamic new up code, I'm just not getting why adding params is so expensive. So I tried one more test...
for (int i = 0; i < 1000000; i++)
{
int j = 5;
string s = "hello";
Color cr = Color.Red;
// 0ms
//int k = j;
// 15ms
//object k1 = j;
// 0ms
//object k2 = s;
// 46ms!!
object k3 = cr;
}
Seems like the boxing & type casting of the int and color is killing the performance. Boxing a string seems to be pretty cheap. The int and color are structs though...
|
|
|
|
|
Boxing and unboxing is notoriously "slow" - although that's a relative term; it's still significantly faster that accessing the disk.
Without knowing the precise details of how you're creating and calling the delegates, it's hard to see whether there's a better alternative.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
The requirements are to create objects on the fly with any number of constructor arguments. Almost like DI, but with a different spin. The only way I thought of being able to do that is to create an object[] as an input parameter and pass the arguments to the constructor as array[0], array[1], array[2] and type casting to the proper type (from ConstructorInfo and ParameterInfo). Hmm... I added support for making constructor params byref, but that didn't help. Also doesn't seem to make a big performance diff if I pass in classes.
|
|
|
|
|
SledgeHammer01 wrote: I need to new up said class as fast as humanly possible.
Do you have any control over how the class/constructor looks? If yes, you could use a delegate as shown here[^].
Bastard Programmer from Hell 
If you can't read my code, try converting it here[^]
|
|
|
|
|
Eddy Vluggen wrote: Do you have any control over how the class/constructor looks? If yes, you could
use a delegate as shown here[^].
Nope. It can be any class.
|
|
|
|
|
Hello all,
Lets say I have a solution (A) or a project (A) which is working properly and its framework is 2.0.
And I have a seperate empty project(B) which is I set its framework to 3.5. When I add my working project(A) to that blank solution what will be my working project's solution? Is it going to be updated to 3.5 automatically from 2.0 to 3.5?
If not, changing project (B)'s(-where project (A) is already been added to project (B) ) framework manually from the solution's property window to 4.0 also changes project (A)'s framework as well(will .Net automatically change that project framework)?
Thanks in advance
|
|
|
|
|
Target framework is a project property and not solution property. It's possible that a solution has different projects with different target framework.
If you have a working project with specific framework and you do not mean to change/extend functionality - do not change target framework...
I'm not questioning your powers of observation; I'm merely remarking upon the paradox of asking a masked man who he is. (V)
|
|
|
|
|
I am using .net framework wcf client and server . The signature of the WCF service interface is simple:
ResponseMessage SendMessage(RequestMessage message)
I create a new type that inherits from "ResponseMessage". This type holds 1 property:
public class TestResponseMessage : ResponseMessage
{
public List<ISomeInterface> Members { get; set; }
}
According to:
http:
http:
It is said that "Members" property should be able to serialize without the need of the [DataContract] on the class and [DataMember] on the property, but it doesn't work - if I don't use it, I get null in the TestResponse.
The reason for asking this is that we've encountered a very strange behavior in our product - sometimes this serialization works and sometimes it doesn't, and it seems to behave differently with same client/server versions on different machines.
Any insights about this will be very welcome.
modified 3-Sep-14 12:07pm.
|
|
|
|
|
I think certain kinds of "advice" without reference to the actual documentation is worthless (as in this case).
If one consults the actual documentation (help) re: "Data Contracts" it notes that:
"All .NET Framework primitive types, such as integers and strings, as well as certain types treated as primitives, such as DateTime and XmlElement, can be serialized with no other preparation and are considered as having default data contracts."
and
"New complex types that you create must have a data contract defined for them to be serializable."
I would say that your "Members" property is NOT a "primitive" type and therefore must have a "data contract".
(The SO example conveniently only shows primitive types).
|
|
|
|
|
Hi,
I am trying to view a photo blob from MySQL on a picture control using the following code:
byte[] rawData = new byte[0];
rawData = (byte[])sql_reader["photo"];
int len = new int();
len = rawData.GetUpperBound(0);
FileStream fs = new FileStream(photo_file, FileMode.OpenOrCreate, FileAccess.Write);
fs.Write(rawData, 0, len);
fs.Close();
curImage = Image.FromFile(photo_file);
pictureEmployee.Image = curImage;
but getting the following error:
Non-negative number required. Parameter name : count
Technology News @ www.JassimRahma.com
|
|
|
|
|
Jassim Rahma wrote: but getting the following error:
On which line?
|
|
|
|
|
no line!
the error shows when I put it in a try-catch block
but If I remove the try-catch I don't get any error but I also don't get the image shown!
Technology News @ www.JassimRahma.com
|
|
|
|
|
You'll still get a line number in the stack track of the exception.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
nothing.
I also noticed the file was not properly created.
is the code ok?
Technology News @ www.JassimRahma.com
|
|
|
|
|
this is the code I am using to create the image and save it to the database:
byte[] employee_photo;
memory_stream = new MemoryStream();
pictureEmployee.Image.Save(memory_stream, System.Drawing.Imaging.ImageFormat.Png);
binary_reader = new BinaryReader(memory_stream);
employee_photo = binary_reader.ReadBytes((int)memory_stream.Length);
binary_reader.Close();
memory_stream.Close();
sql_command.Parameters.AddWithValue("param_photo", employee_photo).MySqlDbType = MySqlDbType.Blob;
Technology News @ www.JassimRahma.com
|
|
|
|
|
You haven't reset the position of the MemoryStream before you try to read it with the BinaryReader, so I suspect you're saving an empty byte array to the database.
Since you're using a MemoryStream, you don't need to worry about the BinaryReader - just use the ToArray method[^] instead:
using (var memoryStream = new MemoryStream())
{
pictureEmployee.Image.Save(memoryStream, System.Drawing.Imaging.ImageFormat.Png);
byte[] employee_photo = memoryStream.ToArray();
sql_command.Parameters.AddWithValue("param_photo", employee_photo).MySqlDbType = MySqlDbType.Blob;
}
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Thanks...
It works now for both insertion and retrieving..
Technology News @ www.JassimRahma.com
|
|
|
|
|
one last question please..
I have a dashboard form and details form.
I am viewing the photo on the dashboard form and when the user clicks edit a details form will be shown with the photo too then user can update the photo from the details form.
I have no problem viewing the photo on the dashboard form.
but when the user clicks edit I get the following error:
The process cannot access the file c:\temp\tmpimage.png because it's being used by another process
How can I avoid this?
Technology News @ www.JassimRahma.com
|
|
|
|
|
Using Image.FromFile locks the file until the application ends. This is a known bug that's been around since .NET 1.0; the workaround is to use the Image.FromStream method instead.
In this case, since you're reading the image from the database, you don't need to store it in a file. Just create a MemoryStream over the byte array returned from the database:
byte[] rawData = (byte[])sql_reader["photo"];
using (var memoryStream = new MemoryStream(rawData))
{
pictureEmployee.Image = Image.FromStream(memoryStream);
}
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
It's not a bug. It's a very unwanted "feature".
The file stays locked for the lifetime of the Image object, not the application. Dispose the Image and the lock goes away.
|
|
|
|
|
Strange - I'm sure it used to keep the image locked until the process ended, but I've just tested in .NET 4.0 and 3.5, and disposing the image does release the lock.
I guess they must have fixed that bug earlier than I thought.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
Great 
Thank you so much
Technology News @ www.JassimRahma.com
|
|
|
|
|
Hi,
i have an sql table which stores times of a machine's 'work start' and 'idle start' times, which is like ;
2014-09-03 12:50:15 start
2014-09-03 13:45:25 idle
2014-09-03 13:56:06 Start
2014-09-03 16:30:35 idle
.
.
and goes like that.
i need to calculate the time between first start and first idle time after the first start, then, time between second start and first idle after the second start, and it goes like that and there are too many lines. at the and i need total work time and total idle time.
Can anyone help me with this?
Thank you.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
