|
I know I can do this with try/catch/finally.
But I really want to know if there is any way to discover if the method I am is actually called by an exception (is inside a catch/finally) or not.
The purpose is to make the method simpler to use, as the caller of the method will not be required to pass the exception (or the boolean) as parameter.
A very ugly example:
using(var someTransaction = new MyTransaction())
{
... code here ...
someTransaction.Commit();
}
And then, in Dispose, I will check:
If the transaction was committed/rollbacked, do nothing.
If the transaction was not committed or rollbacked, but there is no exception, I will throw an exception telling that Commit or Rollback is required.
If there is already an exception, I will rollback and keep the original exception.
- Note: I know how to use try/catch/finally. But I really want to make the use of the code simpler for the others. using keyword looks nicer than try, catch and finally.
modified on Monday, November 23, 2009 12:38 PM
|
|
|
|
|
One possible solution is to have a global variable which all try/catch code can log their exception to. (Probably best to make that a List<exception> and process stuff on or off that exception stack.) Any code can then test the global exception variable to see if there is an exception outstanding. Once you have acted appropriately to that exception then clear it off the stack.
If you have knowledge, let others light their candles at it.
Margaret Fuller (1810 - 1850)
www.JacksonSoft.co.uk
|
|
|
|
|
But this brings the original problem again:
The developer will need to "add/remove" exceptions manually. My question is simple:
.Net has something like that built in (in any class) or not?
I originally thought that I could find catch using StackTrace, but it has no information about catchs.
Is there any way to discover this, even if this is an expensive operation? After all, this will be used only in Debug code.
-> To be honest, I really think that .Net must register exceptions automatically in a "FILO/LIFO" queue and so, when one exception is thrown inside a catch it could automatically fill the "inner exception" property without manual coding. Of course, such list must be ThreadStatic, as one thread can be executing normal code, while the other is executing a catch inside a catch.
|
|
|
|
|
In ASP.NET, there is something like Server.GetLastError();
|
|
|
|
|
Short answer: No.
There are sometimes ways to get an otherwise unhandled exception without a try-catch. In asp.net apps for example you can handle Application_OnError. But it's not a substitute for exception handling elsewhere in the code, only a last-resort chance to at least do things like log errors.
You can use either delegates or polymorphism if you want to standardize exception handling in certain situations. Something like this would be possible:
public static class DbHelper
{
public static RunInTransaction(SqlConnection cnx, Action method)
{
cnx.Open();
SqlTransaction tx = cnx.BeginTransaction();
try
{
method();
tx.Commit();
}
catch (Exception ex)
{
tx.Rollback();
throw new ApplicationException("Error during database transaction.", ex);
}
finally
{
cnx.Close();
}
}
}
Here I used the predefined Action delegate, which is void and takes no parameters, but you could of course
use any delegate you'd like. Or you could use polymorphism: It could be an abstract method that derived classes must implement (obviously the class then can no longer be static!), or the method could accept an object implementing some interface instead of a delegate.
There are some cool possibilities with IDisposable but it is really intended specifically for early release of non-managed resources. It's not meant to be a general try-finally replacement, and certainly not a try-catch-finally replacement.
modified on Monday, November 23, 2009 2:39 PM
|
|
|
|
|
|
Thanks a lot. This solves my problem!
|
|
|
|
|
Hi
I have a rather simple question. I have quite a few try/catch blocks in my system, but I don't really need to handle the exceptions catched. So I just use something like this:
try
{
}
catch (Exception error)
{
}
What I want to know is, since I'm not using "error" that I'm declaring, can I just have a parameterless catch like the following:
try
{
}
catch
{
}
|
|
|
|
|
You can.
But why are you doing it? It's a terrible practice to use exceptions as part of normal program flow. Exceptions are supposed to be *exceptional* - and there are extremely few cases where it would make any sense to just catch any exception, ignore it, and attempt to continue.
|
|
|
|
|
What is the sense in catching an exception and then ignoring it?
|
|
|
|
|
To prevent my system from crashing when an error does occur.
|
|
|
|
|
But once the error has occured, don't you want user to know? Or will your application behave normally even after any kind of exception occurs in it?
50-50-90 rule: Anytime I have a 50-50 chance of getting something right, there's a 90% probability I'll get it wrong...!!
|
|
|
|
|
Presumably your program is supposed to *do* something, not merely "not crash". All that can be achieved by catching exceptions and doing nothing about them is to prevent the program from exiting because of the error. But there's a reason why programs exit by default when errors occur: The program state is no longer well known. Programs close by default in order to prevent a program from *seeming* to work but actually producing corrupt data.
If you are getting exceptions a lot of the time without any associated external reason (hardware failure, network failure, out of disk space/memory, and so on) it just means your code isn't correct, and you should take the time to find out what's wrong with it and correct it.
|
|
|
|
|
Etienne_123 wrote: To prevent my system from crashing when an error does occur
Most often that does not make sense at all.
An exception indicates something went wrong, you can not just ignore it, you should:
1) probably log it
2) act on it (if nothing else, throw it again, so a higher level of the code can act on it).
3) consider informing the user.
And you should only catch the exceptions you can handle in a meaningful way, hence catch them as specific as possible.
Examples:
- when writing to a file, if it fails (error in path, disk full, whatever IOException) you can't ignore that, since then your output would be uncertain; here you must inform the user, unless you have a retry that succeeds.
- when parsing user input, a parse exception has to be reported to the user so he can re-enter data.
The rare exception to the rule is e.g. where you have a retry loop, then all exceptions except for the one in the last retry iteration can be left unhandled and unreported (I suggest to include a comment as to why a catch block is empty then); the exception in the last retry iteration has to be handled though, since that is the exit result of the retry loop.
Luc Pattyn [Forum Guidelines] [My Articles]
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
|
|
|
|
|
I think you need to study exceptions: what they are for and what action you should take when you catch one.
|
|
|
|
|
"On Error Resume Next", and it should be taken out and shot.
I are Troll 
|
|
|
|
|
That one belongs exclusive to VB language
|
|
|
|
|
That exact statement, yes, that's confined to Basic, but the pattern can be found in all languages.
I are Troll
|
|
|
|
|
I use parameterless catches on occasion, such as:
try
{
Utilities.DoSQLNonQuery(cmd);
}
catch (Exception) { }
Which is part of my fault logging system. If I can't log a fault, I can't use the exception, but I don't want to lose the user data either.
No trees were harmed in the sending of this message; however, a significant number of electrons were slightly inconvenienced.
This message is made of fully recyclable Zeros and Ones
"Rumour has it that if you play Microsoft CDs backwards you will hear Satanic messages.Worse still, is that if you play them forwards they will install Windows"
|
|
|
|
|
that is all right, as long as you leave the comment there.
Luc Pattyn [Forum Guidelines] [My Articles]
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
|
|
|
|
|
Yes, you can. But the whole point of catching exceptions is to identify an 'exceptional' situation in your code and handle it properly. Having empty catch blocks is worser than not handling exceptions in your code. As a temporary measure, you can atleast log the error and show a message to the user that something wrong has happened.
|
|
|
|
|
Why would you ask us this instead of just trying it ?
Christian Graus
Driven to the arms of OSX by Vista.
Read my blog to find out how I've worked around bugs in Microsoft tools and frameworks.
|
|
|
|
|
Christian Graus wrote: Why would you ask us this instead of just trying it ?
Come on Christian, you know the answer as well as the rest of us. 
|
|
|
|
|
I did try it and it worked. I was merely making sure.
Why would you reply to a message when you don't aleast try and give an answer? This is the second time you've replied like this to one of my questions. I've said it before and I'll say it again, these forums aren't JUST for experts.
I didn't say I'm not going to handle my exceptions, I just wanted to know whether an empty and parameterless catch will work fine for the time being, as I still need to establish what my user wants to have done when an exception occurs.
|
|
|
|
|
Etienne_123 wrote: I still need to establish what my user wants to have done when an exception occurs.
This is not the way to look at it. If an exception occurs then it indicates a fault in your program, so you need to discover what the fault is and whether the program should continue. There are very few occasions when you should allow the program to continue, as your program, and thus your data/files/security may now be compromised. From this point on anything can happen.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
