CODEWITHSUNDEEP
c#.net-4.5
Lord Relix
Lord Relix

Reputation: 942

How to continue a loop after an error?

I have an application that runs a big loop where it reads data, writes to a PDF and e-mails the files, all in one swoop. Sometimes, an error will appear and I have to retrace back to what caused the error. Basically the whole loop is in a Try/Catch block. There are two loops, basically (in pseudo-code):

try
  // Loop 1
  process 1
    // Loop 2
    process 2
catch
  // Message box error

Is there a way I can continue this loop and just skip the error? Maybe save a log with the exception so I can save it later?

Upvotes: 3

Views: 6361

Answers (3)

pid
pid

Reputation: 11607

Don't use try..catch as a panacea outside the for. Remove it from there and put it around those single instructions that you know may cause trouble.

The tighter they are the better. You will then be able to handle that case without breaking out of the entire loop or procedure:

for(...)
{
    // procedural code

    try {
        // least possible problem code

    } catch(Exception ex) {
        // log/report error 1
        // use continue/break to continue with next cycle or break out of the loop
        Log.Message("Exception (in big loop #01): " + ex.Message);
    }

    // procedural code

    try {
        // least possible problem code

    } catch(Exception ex) {
        // log/report error 2
        // use continue/break to continue with next cycle or break out of the loop
        Log.Message("Exception (in big loop #02):" + ex.Message);
    }

    // procedural code
}

This way you'll also be able to distinguish where the exception occurred.

To log the exception you'll have to implement the Log class this way:

using System.IO;

public class Log
{
    public static void Message(string message)
    {
        using (StreamWriter writer = File.AppendText("path_to_dir\\log.txt"))
        {
            writer.WriteLine(message);
        }
    }
}

This Log class is very very basic and can be improved in following ways:

  • don't use static methods
  • don't use singleton, inject a logger object into the big-loop-function
  • add time and date in front of each line
  • add standard facility and severity (best if taken from syslogd)
  • allow to also send email if severity is above threshold level
  • allow for log rolling based on time, lines or file size
  • what happens if an exception occurs while logging an exception? a meta-exception!?
  • much more

What I put here is just to give you an idea about logging errors. It is a quite wide area of expertise.

Upvotes: 4

ciamej
ciamej

Reputation: 7068

Maybe like this?

// Loop 1
  try
    process 1
    // Loop 2
      process 2
  catch
    // Message box error

It depends on what exactly you want. What kind of errors do you want to ignore, those stemming from the PROCESS ONE, PROCESS TWO or both? The pseudocode above will ignore all the errors and continue the outer loop 1.

Upvotes: 2

rae1
rae1

Reputation: 6144

Just use another try ... catch inside the loop,

try
  foreach () // Loop one
    try
      foreach () // Loop two
    catch 
      // Log error, or ignore, then it continues loop 1    
catch
  // Message box error

You can nest as many try ... catch constructs as needed; however, this will increase the complexity of your program quite rapidly, so use with caution; and when possible, isolate blocks of code to functions/method, each with it is own, required recovery procedure.

Upvotes: 5

Related Questions