Visual Basic

Re: Global Try..Catch block again

15-Feb-08 10:29

Well I've discovered I still dont' know where to put a global try...catch block in vb.net. I figured it would go in the Mainform.mybaseload method but that doesn't seem to do it.

Thanks.

Luc Pattyn15-Feb-08 11:12

15-Feb-08 11:12

Hi again,

this is how it is done in C#:

static class Program {
	/// The main entry point for the application.
	[STAThread]
	static void Main() {
		Application.EnableVisualStyles();
		Application.SetCompatibleTextRenderingDefault(false);
		try {
			Application.Run(new myMainForm());
		} catch(Exception exc) {
			log(exc.ToString());
		}
	}
}

The above code instantiates and executes the main form class.

For a regular Windows app, VB.NET seems not to rely on a Main() method, but directly
launches the startup object.
However I trust in a VB.NET module you could do exactly what the C# code does.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

This month's tips:
- before you ask a question here, search CodeProject, then Google;
- the quality and detail of your question reflects on the effectiveness of the help you are likely to get;
- use PRE tags to preserve formatting when showing multi-line code snippets.

Disable Try/Catch

cstrader23215-Feb-08 6:57

15-Feb-08 6:57

Can I disable my try..catch block handling while debugging? Thanks

Dave Kreskowiak15-Feb-08 7:14

15-Feb-08 7:14

Nope. It's part of your code. You'll have to comment it out before you start the debugger.

A guide to posting questions on CodeProject[^]

Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007

cstrader23215-Feb-08 7:17

Re: Disable Try/Catch [modified]

15-Feb-08 7:17

Too bad..that's why making too many of them makes it difficult to debug.

Dave Kreskowiak15-Feb-08 7:20

15-Feb-08 7:20

Not really. It's how you break down your problems and what you put in the Try/Catch block that really matters. (Hint, hint)

Also, it helps if you're not putting the bulk of your method code in the Try block. Keep the Try/Catch blocks small, encompasing only very small operations. Say you have a method that queries a database, then loops through the returned recordset and sends an email to every address it finds. There's a ton of places this can fall on its face. Do you put all of this in a single Try/Catch block?? Nope.

Really, this can be broken down into a bunch more methods, each handling very small parts of the process. Creating and returning a connection to the database is one, exectuing an arbitrary query is another, iterating over a dataset, pulling email addresses out, sending an email to an address, ... These are all seperate methods, each with it's own Try/Catch opportunities. The more atomic the operation of each method, the better.

Luc just nailed it in his post here[^].

A guide to posting questions on CodeProject[^]

Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007

modified on Friday, February 15, 2008 1:27 PM

Luc Pattyn15-Feb-08 7:48

15-Feb-08 7:48

In my experience, it is not the number of the try-catches that poses a problem;
a lot of people seem to stuff the catch blocks with MessageBox.Show which is a
terrible idea, since now you get tired of reading messages, one at a time, and having
to close them over and over again.

I prefer to provide a simple "log" method that takes a single string input, and
logs it somehow. I then implement the log based on Console.WriteLine, a file operation,
a listbox operation, or whatever suits my needs at the time. The advantages are:
- I don't have to interact with this logging
- I get it all sequentially (I tend to automatically add a timestamp too)
- I can save it, and compare different runs

I often temporarily add a ListBox to my main form, just for showing the logging.
MessageBox really is not my prefered debugging tool!

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

cstrader23215-Feb-08 7:59

15-Feb-08 7:59

What I'm mostly having trouble with is determining where the error is, not the error handling itself. When my method is big, I can't tell where the error occurs within it, when it's running from the .exe on a different machine.

I don't seem to be getting any line numbers to help out. Perhaps I'm not in debug mode, although the .exe is written to the debug directory.

Thanks again.

Luc Pattyn15-Feb-08 8:03

15-Feb-08 8:03

strange, I just did a little VB app (VS2005 VB.NET Express) with a deliberate error,
and it exceptioned with the correct line number right away.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

Dave Kreskowiak15-Feb-08 8:24

15-Feb-08 8:24

cstrader232 wrote:
not the error handling itself. When my method is big, I can't tell where the error occurs within it

This is a hint that your method is too big and needs to be broken down into smaller pieces. Also, it sounds like you're putting way too much inside the Try block. A Try block should encompass as few lines as possible that can fail. You don't wrap an entire method. You wrap small sections where each section has a specific task.

A guide to posting questions on CodeProject[^]

Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007

Luc Pattyn15-Feb-08 10:18

15-Feb-08 10:18

Hi Dave,

I beg to differ on this one, I would put as much code in a try block as logic dictates.

If my method needs 100 simple steps without any side-effects, and I don't care where
it fails, I'll put them all in a single try.
The net result would be either success, or failure in which case the details
of the failure are (should be) apparent from the exception, and probably the catch
block would contain code to throw an application-specific new exception (and store the
original as an inner exception).

If on the other hand, the steps have lots of side-effects (allocating resources,
opening files, etc) I would use more try/catch/finally blocks, even nested, to
handle this elegantly.

But I would avoid the fully sequential try one step, catch, try next step, ... approach
since then there could well be more error handling code than straight code, yielding lower
programmer efficiency, more code and more execution paths to test.

I do stress the use of line numbers while debugging though, a simple technique a lot
of people seem unfamiliar with (the fact that the Visual Editor by default does not
show them looks like a mistake to me).

Regards,

Luc Pattyn [Forum Guidelines] [My Articles]

Dave Kreskowiak15-Feb-08 10:42

15-Feb-08 10:42

Luc Pattyn wrote:
I beg to differ on this one, I would put as much code in a try block as logic dictates.

If my method needs 100 simple steps without any side-effects, and I don't care where
it fails, I'll put them all in a single try.

True. I guess I tailored my answer to his case. If he can't find where the offending code is, that's a sign he's got too much in the Try block.

I try to keepo me Try's as small as possible, but it, agreeably, depends on the situation.

Luc Pattyn wrote:
But I would avoid the fully sequential try one step, catch, try next step, ... approach
since then there could well be more error handling code than straight code, yielding lower
programmer efficiency, more code and more execution paths to test.

I think I should clarify that I'm not saying he should break each tiny step into it's own Try/Catch. I am saying that it should be closer to wrapping each step that makes up a transaction.

For instance, depending on the situation and granularity of the error handling in the requirements, wrapping creating and opening a connection object and running a query and returning a result set is probably going to be more "friendly" than wrapping each step in that same code. If you're going after a generic DAL library that executes arbitrary SQL on a database, then you might want to wrap each of those steps. It just depends on the flexibility or importance of the transaction in question.

Some transactions are tolerant of a single item failure and continue on. Others need to know so they can stop at a certain point. And still others need to backout in the event of a certain set of failures, but can continue on in the event of a smaller failure.

I've probably done a piss-poor job of explaining this. I apologize in advance. It's late in the day and I'm starting to fall asl..... Sleepy | :zzz:

A guide to posting questions on CodeProject[^]

Dave Kreskowiak
Microsoft MVP
Visual Developer - Visual Basic
2006, 2007

Luc Pattyn15-Feb-08 11:01

15-Feb-08 11:01

Dave Kreskowiak wrote:
I've probably done a piss-poor job of explaining this

If so, you have amply rectified it. Big Grin | :-D

Dave Kreskowiak wrote:
It's late in the day

That's all relative.
Gnite.

Luc Pattyn [Forum Guidelines] [My Articles]

Exception information detail

Paul Conrad15-Feb-08 12:43

Paul Conrad

15-Feb-08 12:43

Very good points there, Luc.

"The clue train passed his station without stopping." - John Simmons / outlaw programmer

"Real programmers just throw a bunch of 1s and 0s at the computer to see what sticks" - Pete O'Hanlon

cstrader23215-Feb-08 5:27

15-Feb-08 5:27

Does an exception give any information more specific than the method in which it occurs and the exception type and message? If you have a long method it is sometimes hard to see where the error is.

Thanks!

Luc Pattyn15-Feb-08 6:05

15-Feb-08 6:05

Hi,

an Exception holds a lot of information:
- the Message (a one-line string)
- extra information (such as the file path on a file I/O error)
- the Stack TraceBack containing a list of methods, with line numbers when available
- and the same for any inner exception.

To get it all you need to use ToString(). Just looking at Message won't help much.

To locate the problem line, use a debug build and look at the very first line number
that appears in the stack traceback.

BTW: you can teach Visual to always show line numbers in the editor, see Tools/Options/
TextEditor/...

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

cstrader23215-Feb-08 6:22

15-Feb-08 6:22

Thanks Luc!!... might I ask another question of you?... what's the best way to capture the exception within the entire project? I have specific try..catches, but I don't want to make a separate one for each method.

cstrader23215-Feb-08 6:32

15-Feb-08 6:32

I'm not seeing line numbers in the stacktrace. How can I be sure I'm compiling in debug mode?

Thanks

Luc Pattyn15-Feb-08 7:30