|
New job, new surprises. This is everywhere in the C# apps here:
MyObject myObject = new MyObject();
if (!myObject.doSomething())
{
LogError(myObject.ErrMsg);
}
doSomething returns false when it encounters an error and saves the error message in the public property ErrMsg, which the caller will use to see the error.
What's wrong with that? Well I must admit I don't really know but it looks wrong to me. Isn't this what exceptions are for:
MyObject myObject = new MyObject();
try{
myObject.doSomething();
}
catch(Exception e){
LogError(e.Message);
}
When I mentioned it I was told throwing exceptions was too expensive. They do have some try/catch blocks but only for exceptions thrown by the .Net framework itself. All application errors are handled with this (at least for me) odd ErrMsg property.
How expensive is it to throw exceptions really? Is it reason enough to come up with a different error handling mechanism? Is it what they are doing such a bad thing that I should push to change it?
Cheers.
|
|
|
|
|
http://blogs.gotdotnet.com/cbrumme/commentview.aspx/d5fbb311-0c95-46ac-9c46-8f8c0e6ae561[^]
Short version, the CLR will always carry the baggage required to allow exceptions, and the cost of Exceptions is only high if they are likely to be thrown frequently inside tight loops. Otherwise, the potential cost of someone deciding they don't need to check the error code somewhere is much higher.
And seeing as you've just started, you should probably read that entire article, and present the most solid case you can, to impress your new employer.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Thanks for that Christian. Heavy reading for me, but interesting.
"the potential cost of someone deciding they don't need to check the error code somewhere is much higher"
That's my problem with c#'s unchecked exceptions. You can also forget to catch exceptions (or just be lazy and not do it), just like you could forget to check for error codes. I understood the argument that exceptions are only expensive when you, well, throw them, but I just don't see how unchecked exceptions help in not ignoring errors.
From the article: "It’s relatively easy to forget to check for a returned error code. It’s much harder to inadvertently swallow an exception without handling it".
I don't understand that statement. It might be hard to "inadvertently swallow an exception" but it's way too easy to ignore it (specially when you don't even know what exceptions a method could throw!). And that's just as bad as not checking for error codes ...
I want to present a case in favor of exceptions to my employer but this unchecked exceptions are giving me headaches. As you can guess by now I have a Java background ...
Cheers
|
|
|
|
|
elguaro wrote:
"It’s relatively easy to forget to check for a returned error code. It’s much harder to inadvertently swallow an exception without handling it".
I think what's meant here is that forgetting to check an error code is as easy as to forget to catch an exceptions, but both have different consequences. Whereas the unchecked error code just gets lost, the CLR will popup that an unhandled exception occured in your app.
So it's easy to forget the exception handling, but it's much easier to detect it.
www.troschuetz.de
|
|
|
|
|
elguaro wrote:
You can also forget to catch exceptions (or just be lazy and not do it), just like you could forget to check for error codes. I understood the argument that exceptions are only expensive when you, well, throw them, but I just don't see how unchecked exceptions help in not ignoring errors.
Because an error code ignored does nothing, an exception ignored brings your app to a halt.
elguaro wrote:
It might be hard to "inadvertently swallow an exception" but it's way too easy to ignore it (specially when you don't even know what exceptions a method could throw!).
You shouldn't ever catch Exception, only the Exception types that you anticipate and know how to deal with. If others get thrown, add them and work out how to deal with them.
elguaro wrote:
As you can guess by now I have a Java background ...
LOL - no, I didn't see that coming at all
The core issue is that if you don't catch an exception, you'll find out about it, and you'll be forced to write code to handle it. Consider exception handling to be error code checking with built in unit tests 
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Christian Graus wrote:
Because an error code ignored does nothing, an exception ignored brings your app to a halt.
That's if the exception is actually thrown. If I call a method and I never get an exception while developing I could forget to catch exceptions. You go whislting to prod and bang! Another unhappy user.
Christian Graus wrote:
If others get thrown, add them and work out how to deal with them.
Ok, so the deal is you add them as you find them. Again, you might only find one particular exception in prod. Had you been forced to handle it maybe something could have been done about it.
Christian Graus wrote:
LOL - no, I didn't see that coming at all
Yea that's why I thought I clarify it ... 
Anyway, I still think what they're doing here is wrong even with unchecked exceptions ...
Thanks for you comments ...
Cheers
|
|
|
|
|
elguaro wrote:
If I call a method and I never get an exception while developing I could forget to catch exceptions. You go whislting to prod and bang! Another unhappy user.
Well, that's what we like to call 'the real world' Any problem that doesn't come up in your obviously extensive testing/unit tests is going to be esoteric at best, and it's still better that you find out about it. There's nothing harder to diagnose than an error that only occurs on a client machine. With an exception, you'll actually know where the error is, instead of trying to diagnose ripple effects.
elguaro wrote:
Again, you might only find one particular exception in prod.
You do testing, right ? Most of the time, if you think about what you're asking a function to do, and/or read the docs, you should have a fair idea what exceptions are likely to be thrown.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Well we obviously work in very different environments. I have 3 years working in the Government (in Australia this is) and documentation is not that good. It doesn't exists really. And from what I've seen, developers do forget to catch exceptions or simply choose to ignore them. These are people
who are learning OO with C# and if the compiler doesn't force you to catch the exception you just don't do it. Even if you do your testing and the testing team do their thing afterwards, for some strange reason in my "real world" no matter how much testing you do, you can only expect the unexcepted in production. Then you get a stupid error that you could have had a chance of recovering from had you caught it. 
Christian Graus wrote:
Any problem that doesn't come up in your obviously extensive testing/unit tests is going to be esoteric at best, and it's still better that you find out about it
Frustrations aside  , just knowing that an error occurred is not good enough when you could have dealt with it and (why not!) recover. And no matter how extensive your testing is, sh*t will happen. You should assume it will too.
Christian Graus wrote:
if you think about what you're asking a function to do, and/or read the docs, you should have a fair idea what exceptions are likely to be thrown
Disagree. You shoudn't try to infer what exceptions are likely to be thrown just by looking at a method's name. You'd be implying implementation details.
|
|
|
|
|
elguaro wrote:
developers do forget to catch exceptions or simply choose to ignore them.
You are missing my point. If I choose to catch Exception e, then I will swallow all exceptions, this is an ACTIVE error I make. If I don't put a try/catch at all, then the program will *crash* because I forgot to catch an exception. Simple as that.
elguaro wrote:
for some strange reason in my "real world" no matter how much testing you do, you can only expect the unexcepted in production.
That's perhaps a sign that the test cases need to be broadened Even then, it's inevitable that some issues will be found by the client, that's just life. As I said, how much better for them to get an exception message that tells you exactly what went wrong on a production machine, rather than a general crash that leaves you spending hours looking at the code, trying to figure out how on earth the app got into a state so it did what they reported ?
elguaro wrote:
just knowing that an error occurred is not good enough when you could have dealt with it
But again, that's the point. We should be able to tell what errors are likely to occur in the majority of cases.
elguaro wrote:
You shoudn't try to infer what exceptions are likely to be thrown just by looking at a method's name.
Sorry, that's not what I meant. I meant that if I, within a function, try to open a file for reading (for example), I should be able to work out what exceptions are likely to occur within that API call, and therefore what exceptions I should be catching.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Christian Graus wrote:
If I don't put a try/catch at all, then the program will *crash* because I forgot to catch an exception
You wouldn't forget if the compiler forced you to put in that try/catch. And that's one crash less in your system.
Christian Graus wrote:
how much better for them to get an exception message that tells you exactly what went wrong on a production machine, rather than a general crash that leaves you spending hours looking at the code
How much better not to have an error in the first place? Maybe the error was something you could have recovered from but forgot to put that try/catch.
Of course having unchecked exceptions is better than not having anything at all. What I'm saying is I'd prefer to have checked exceptions.
We could spend our lifes and never agree on this one ...
Thanks for the discussion ...
|
|
|
|
|
|
Given that DX9 is an utter joke, I'm using the Media Player control to play video files in my app. The thing is, I want to build a queue of files in my code, and play them, but it seems to do that, I need to load the file, wait for the player state to be set to ready twice, then I can play. This I did, and it works fine on my machines, but not on the clients. Does anyone have any info or know of a good source of info on maintaining a file queue using the WMP SDK ?
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Probably not what you're looking for... WMP is so bloated we took a different approach. We wrapped the MCI calls in C# and maintain our own queueing mechanism. Our "player" is event driven once started and runs on a constantly filling queue which pretty much runs all the time (well, 20 hours a day). An event tells us when the end of the queue is near (3 items) and we re-fill. Each notification from our video component drives the loading and playing of the next media item. There's tons of examples out there for using MCI directly. Sorry, but I don't have alot of WMP experience myself.
|
|
|
|
|
Christian Graus wrote:
Given that DX9 is an utter joke,
Considering that DirectX was written as a fast interface between games and video cards, and not a full featured media player, it's no suprise that it's not doing the job you want it to do.
The Windows Media Player control will do the job, but it's doesn't have a very friendly developer interface.
The best method would be to write your own component using Windows MultiMedia API's. This will give you the best control with the greatest flexibility, but also will force you to write a large amount of code.
RageInTheMachine9532
"...a pungent, ghastly, stinky piece of cheese!" -- The Roaming Gnome
|
|
|
|
|
Dave Kreskowiak wrote:
not a full featured media player, it's no suprise that it's not doing the job you want it to do.
Well, actually, if the two classes in the AudioVideoPlayback namespace worked at all, then they would be perfect. I don't need much, just for videos to play, and a callback when they end. I've written an article on the site if you're interested to know all the ways in which this utterly fails to work.
Dave Kreskowiak wrote:
The Windows Media Player control will do the job, but it's doesn't have a very friendly developer interface.
So I have discovered
Dave Kreskowiak wrote:
The best method would be to write your own component using Windows MultiMedia API's. This will give you the best control with the greatest flexibility, but also will force you to write a large amount of code.
I just may spend the weekend doing that. I didn't realise it was an option, that is to say, I did not know they existed. I'll check it out.
thanks.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
I'm taking over an application from a developer that just left the company. The application is mostly "completed" with just a few bugs and a couple pieces of functionality I'll have to add (no big deal). It's really just a datagrid tied to a sequel server through a proper dataset. The user has several types of searches possible and for each one there is a different TableStyle used in the DataGrid but in any case the same dataset is used.
My problem is that they want to sort by clicking on the column headers. I know I can turn that on easy enough (and I have). However, it doesn't "appear" that the DataTable being bound to the dataset of the datagrid is being reordered with the dataview of the datagrid (too many freagin data's). Anyway, I know I could do a Select() against the DataSet() using the "MappingName" of the Column Header. This SHOULD give me the same ordering that the grid control is giving me, right?
Is there a way to automate this? The reason I need the grid's "view" and the datatable to stay in Sync (order wise) is if a user-double clicks a row it opens an editor dialog with alot of custom options that wouldn't really work by in-line editing. Also, if the user sorts by the column header I want to be able to retain the proper selected record (if one was selected) once the list is redrawn.
Any help? I can probably hack something but I'd really like to do it properly. I'm a bit new to using this control and, in fact, Datasets, etc.
|
|
|
|
|
Is there a way in C# to specify the hostname in a sql connection string. I know it can be done in C++, but I have searched the documentation for c# and it doesn't appear to be an option.
I am trying to replace the hostname (which by default is the machine name) with an app name so when I do an SP_WHO I can see exactly what app is responsible for what connection.
Darryl Borden
Principal IT Analyst
darryl.borden@elpaso.com
|
|
|
|
|
It has nothing to do with C# - merely the ADO.NET provider (or OLE DB provider, if using the System.Data.OleDb namespace members).
SQL Server supports a hostname. This is the "Data Source" property, where "Initial Catalog" is the database on that RDBMS.
Example:
string connstr = string.Format("Data Source={0}; Initial Catalog=MyDB; " +
"Integrated Security=SSPI", hostname);
using (SqlConnection = new SqlConnection(connstr))
{
}
This posting is provided "AS IS" with no warranties, and confers no rights.
Software Design Engineer
Developer Division Sustained Engineering
Microsoft
[My Articles] [My Blog]
|
|
|
|
|
Hi,
Atlast I have implemented my program to convert an array of Pixel data.
step1)- first of all My original image is a 1650x2544 (widthxheight) and i have stored all the pixel data into an array of (2544x1650).
step2)- I have created an empty Bitmap of respective height and width.
step3)- Now i copied respective pixel data fromt the array data into all the 3 planes (R,G,B). as i am creating a monochrome image i am copying the same value for each three R,G,B values.
step4)- returning the image...
can anyone tell me whether wht i am doin is correct...
here is the code...
private void File_Open(object sender, System.EventArgs e)
{
OpenFileDialog openFileDialog = new OpenFileDialog();
openFileDialog.InitialDirectory = "c:\\" ;
openFileDialog.RestoreDirectory = true ;
int[,] PixelData;
int Height=0, Width=0, j;
if(DialogResult.OK == openFileDialog.ShowDialog())
{
ChosenFile=openFileDialog.FileName;
PixelData = ConvertFile(ref Height, ref Width);
m_Bitmap =new Bitmap(Width,Height,PixelFormat.Format24bppRgb);
if (Display (m_Bitmap, PixelData))
{
this.AutoScroll = true;
this.AutoScrollMinSize = new Size ((int)(m_Bitmap.Width * Zoom), (int)(m_Bitmap.Height * Zoom));
this.Invalidate();
}
}
}
int[,] ConvertFile(ref int H, ref int W)
{
int Height=0, Width=0, RGBWidth=0, Value=0, i;
int[] LUT = new int[4096];
FileStream InStream = new FileStream(ChosenFile, FileMode.Open, FileAccess.Read);
long nBytesToRead = InStream.Length;
byte[] ByteArray = new byte[2];
for ( i=0; i<256; ++i)
{
LUT[j]=LUT[j+1]=LUT[j+2]=LUT[j+3]=LUT[j+4]=LUT[j+5]=LUT[j+6]=LUT[j+7]=LUT[j+8]=
LUT[j+9]=LUT[j+10]=LUT[j+11]=LUT[j+12]=LUT[j+13]=LUT[j+14]=LUT[j+15]=i;
j=j+16;
}
for (i=0; i<nbytestoread; i+="2)
{"
int="" nextbyte="(int)" instream.read(bytearray,="" 0,="" 2);
="" if="" (="" i="">=4 && i<=7)
{
if ( i==4)
Height = (int)BinaryToDecimal(ByteArray);
else
Width = (int)(BinaryToDecimal(ByteArray))/2;
}
else if ( i>=128 )
{
break;
}
}
H = Height;
W = Width;
RGBWidth = Width*3;
int[,] PixelData = new int[Height,Width];
for (int x=0; x<height; ++x)
{
for="" (int="" y="0;" y<width;="" ++y)
{
int="" nextbyte="(int)" instream.read(bytearray,="" 0,="" 2);
value="LUT[BinaryToDecimal(ByteArray)];
" pixeldata[x,y]="Value;
i" +="2;
if" (i="">=nBytesToRead)
break;
}
if (i>=nBytesToRead)
break;
}
return PixelData;
}
public static bool Display(Bitmap b, int[,] PixelData)
{
BitmapData bmData = b.LockBits(new Rectangle(0, 0, b.Width, b.Height), ImageLockMode.ReadWrite, PixelFormat.Format24bppRgb);
int stride = bmData.Stride;
System.IntPtr Scan0 = bmData.Scan0;
int j;
unsafe
{
byte * p = (byte *)(void *)Scan0;
int nOffset = stride - b.Width*3;
for ( int y=0; y< b.Height; ++y)
{
for ( int x=0; x< b.Width; ++x)
{
p[0] = (byte)(255-PixelData[y,x]);
p[1] = (byte)(255-PixelData[y,x]);
p[2] = (byte)(255-PixelData[y,x]); p += 3;
}
p += nOffset;
}
}
b.UnlockBits(bmData);
GrayScale (b);
return true;
}
But I am not getting the image tht i am supposed to get...
thanks in advance,
Suman
|
|
|
|
|
Read my image processing articles in the C# Multimedia section to see how you can access bytes within the image without the copy stage you're taking.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
Hi Graus,...
hmm... atlast I succeded in doin it... and i am able to view the image...and save it as a bitmap..
thanks alot for ur support and advices and sorry for disturbing u so many times on this issue...
ok then catch u later over my project doubts...
Suman
|
|
|
|
|
Hi Graus,
I am trying to plot the histogram of an image.... i am able to calculate all the values required to plot the histogram... but i dont know how to plot it in C#... can u advise me wht i have to do to pop-up a window and show the plot in it ( i mean to plot the graph in it)....
Thanks in advance,
Suman
|
|
|
|
|
My boss has given me an order to display our word documents in PDF format.. Anyway of doing this in C##?
Any advice would be appreciated..
|
|
|
|
|
I believe there's a couple of .NET pdf tools that advertise on the site. There's nothing built in.
Christian
I have several lifelong friends that are New Yorkers but I have always gravitated toward the weirdo's. - Richard Stringer
|
|
|
|
|
i am developing an image processing application as an assignment. i need help. how to rotate an image to any degree from 0 to 360. i have used the buildin method
new Bitmap(fileDialog.FileName).RotateFlip(RotateFlipType.Rotate180FlipNone);
but these r limited options. further more it would be great if unsafe code would be used.
Ahmad Zafar
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
