Trivial Catch block question

In a test harness I wanted to put some debugging output to System.out and keep one line per thread execution so I could read it more easily.
I have a try catch block, and in both sections my debugging prints out with System.out.print, so they should appear on the same line. The strange thing I have noticed is that when ever the catch block is entered the code prints on a new line and I dont know why.
Below is a snippet of code, its pretty simple but illustrates this:
try {
   System.out.println("starting test - ");
   //at this line the test code runs
   System.out.println("success");  
} catch(Exception e) {
   System.out.println("fail");
}The output for a successful test is like:
starting test success
And for a fail:
starting test -
fail
I am just curious to know the answer, its like a new line character is output when the print statement is within the try catch block.

MarksmanKen wrote:
You want to use System.out.print not System.out.println
Notice the difference?
One automatically starts a new line and one just carrys on the current line.Wrong: One automatically ends the current line, while the other one does not!
A subtle difference, but important:
This code will print "foo" and "bar" on two separate lines:
System.out.println("foo");
System.out.print("bar");This code will print "foobar" on one line:
System.out.print("foo");
System.out.println("bar");

Similar Messages

  • How to get the returned error messages in the Try/Catch block in DS 3.0?

    A customer sent me the following questions when he tried to implement custom error handling in DS 3.0. I could only find the function "smtp_to" can return the last few lines of trace or error log file but this is not what he wants. Does anyone know the answers? Thanks!
    I am trying to implement the Try/Catch for error handling, but I have
    hard time to get the return the msg from DI, so I can write it to out
    custom log table.
    Can you tell me or point me to sample code that can do this, also, can
    you tell me which tables capture these info if I want to query it from
    DI system tables

    Hi Larry,
    In Data Services XI 3.1 (GAd yesterday) we made several enhancements for our Try/Catch blocks. One of them is the additional of functions to get details on the error that was catched :
    - error_message() Returns the error message of the caught exception
    - error_number() Returns the error number of the caught exception
    - error_timestamp() Returns the timestamp of the caught exception.
    - error_context() Returns the context of the caught exception. For example, "|Session Datapreview_job|Dataflow debug_DataFlow|Transform Debug"
    In previous versions, the only thing you could do was in the mail_to function specify the number of lines you want to include from the error_log, which would send the error_log details in the body of the mail.
    Thanks,
    Ben.

  • Can error handling framework use fault policies and BPEL catch blocks

    As part of an integrated customer solution that makes substantial use of SOA Suite (including OSB, BPEL, Mediator, etc) we are in the process of designing an “Error Hospital” subsystem to expose a web service API specifically for BPEL services and/or the OSB layer to call whenever they encounter an exception which they cannot handle so that the appropriate user is notified of the error (via email) and the error is recorded (in an appropriate common fault format, with error codes elaborated into more meaningful descriptions, etc) within OEM, immediately prior to returning the exception to the caller.
    The problem we have experienced in our design is as follows:
    •     We have an error handling service to be called, inside the BPEL catch blocks, when an error occurs to notify the appropriate user, via email, that an error has occurred;
    •     We also use the error handling XML fault-policies so that we can recover from an error using OEM (via human intervention);
    •     The error service needs to be called before the process is suspended (waiting for human intervention) so the email notification and error message formatting occurs BEFORE the recovery action. Unfortunately the fault policies trigger BEFORE the error handler is able to transform the error into the appropriate format and notify the user so the error recorded is not the reformatted one and the user is never notified.
    The key question is how can a BPEL, upon catching an exception, invoke the error handling service and AFTER that send the error to OEM (and have the process suspended).
    It seems the two mechanisms "cannot play nicely together"

    As part of an integrated customer solution that makes substantial use of SOA Suite (including OSB, BPEL, Mediator, etc) we are in the process of designing an “Error Hospital” subsystem to expose a web service API specifically for BPEL services and/or the OSB layer to call whenever they encounter an exception which they cannot handle so that the appropriate user is notified of the error (via email) and the error is recorded (in an appropriate common fault format, with error codes elaborated into more meaningful descriptions, etc) within OEM, immediately prior to returning the exception to the caller.
    The problem we have experienced in our design is as follows:
    •     We have an error handling service to be called, inside the BPEL catch blocks, when an error occurs to notify the appropriate user, via email, that an error has occurred;
    •     We also use the error handling XML fault-policies so that we can recover from an error using OEM (via human intervention);
    •     The error service needs to be called before the process is suspended (waiting for human intervention) so the email notification and error message formatting occurs BEFORE the recovery action. Unfortunately the fault policies trigger BEFORE the error handler is able to transform the error into the appropriate format and notify the user so the error recorded is not the reformatted one and the user is never notified.
    The key question is how can a BPEL, upon catching an exception, invoke the error handling service and AFTER that send the error to OEM (and have the process suspended).
    It seems the two mechanisms "cannot play nicely together"

  • Fault Handling in BPEL - Continue after Catch Block

    Hi,
    I have a catch block in my BPEL process.
    On exception, the catch block catches the error and terminates the BPEL Process.
    Is it possible to loop back the flow after the catch sequence?
    Thanks,
    Venkat

    Hi,
    Thanks for your reply
    Sorry, I was not clear with my question.
    I have a infinite loop in my process. Inside the loop I have an invoke operation(Invokes another BPEL process) and a wait operation. If any exception occurs, my BPEL process terminates. However, I wanted my BPEL process to just continue in the loop.
    I solved the issue with the help of this How to handle fault when BPEL process is polling a FTP server which is down
    i.e, using the Scope activity. Place my invoke operation inside a scope and put a catchAll block only for that scope. Its working fine
    Thanks,
    Venkat

  • Return statement at the end of try or after catch blocks

    Hi
    Can anyone tell me which is the better practice - to put the return statement at the end of try block or after all the catch blocks ie at the end of method.
    Eg
    Method A()
    String str;
    try{
    str= [some code]
    return str
    catch(Exception e)
    System.out.println("Exception");
    } // end of method
    OR
    Method B()
    String str;
    try{
    str= [some code]
    }catch(Exception e)
    System.out.println("Exception");
    return str
    } // end of method

    I always try to work with only one exit point for each method.
    For readability I always put the return statement as close to the end of the method as possible.
    In this particular question,
    I think you should put the return at the end of the method (for readability, since this is what you are
    familiar with), but when I have a try-catch clause, I usually have an unrecoverable error and
    should throw this further down the tree, so it usually becomes
    try{
       return   
    }catch(...){
       // write out some logging information
       rethrow exception or throw another exception
    }I think you should NEVER reflect the occurence of an error in the return-value of the
    method when an unrecoverable exception occurs. Just rethrow this exception or throw another method.
    Other methods look like
       Object result = new...
       return result;As for the specific case of a repetitive if-case:
    This is possible in two versions:
    With one return and a result-object
    Object result = new ...
    if(..)
       result = ...
    else if(..)
       result = ...
    return result;With every time a return
    if(...)
       return ...;
    else if(...)
       return ...;For me the above two possibilities make no difference,
    but I find the second version (which I hated when I began programming)
    to be somewhat more of a (self-made) standard nowadays.
    But I do not think this particular case makes much of a difference.
    kind regards,

  • Performance impact on using too much try catch block

    I have several questions here:
    1. The system that I'm developing requires to be high performance, but I am not sure how will try catch block affect overall performance.
    2. I wanted to know which would be more efficient (result in faster processing)
    Have several generic try catch OR catch all exceptions individually?
    ex:
    try {
    } catch (Exception e){
    }vs.
    try{
    } catch (MalformedUrlException me){
    } catch(SQLException){
    }3. Which one would be faster, one big try catch block or several small try catch blocks?
    ex.
    try{
    //read from io file
    //query database
    //parse data
    //write to file
    //query database again
    } catch(Exception e){
    //log exception
    }vs.
    try{
    //read from io file
    } catch(FileNotFoundException fnfe){
    //log exception
    try{
    //query database
    }catch(SQLException se){
    //log exception
    try{
    //parse data
    }catch(SaxParserException saxe){
    //log exception
    try{
    //query database again
    }catch(SQLException se2){
    //log exception
    try{
    //write to file
    } catch(FileNotFoundException fnfe){
    //log exception

    1. The system that I'm developing requires to be high performance, but I am not sure how will try catch block affect overall performance.Compared to what? You can't write an equivalent program that doesn't have a try-catch block, so the answer would have to be that it doesn't affect performance at all.
    2. I wanted to know which would be more efficient (result in faster processing)Have several generic try catch OR catch all exceptions individually?
    You still have it backwards. Do you need to do different things for different exceptions? If so, then that's what you have to do and there is no other code that might be "faster".
    Here's what you should do. Write the code that needs to be written. Don't leave out necessary stuff because of performance reasons. (If you left out all your code, the program would run much faster.) Then find out which parts of the program ACTUALLY take the most time and work on speeding them up.

  • Try Catch Blocks

    This may be too vague of a question but is it bad practice to use one large try catch block? In my mind it seems like it would produce "cleaner" code as far as programmer reading is concerned, but does it affect the program adversely?
    try {
    // code goes here
    } catch (somekindofexception e){
    // print exception e
    } catch (anotherkindofexception e){
    // print exception e
    } // continue for all possible exceptions

    bolivartech wrote:
    I was mistaken it looks like the particular function I was thinking of throws 3 kinds of RuntimeExceptions, but the same idea applies right?
    try {
    } catch (RuntimeException e) {
    System.err.println("Caught RuntimeException: " +  e.getMessage());
    Yes, you can do that if you want to handle all of them the same.
    However, if you want to handle different exceptions differently, you'd catch those specific exceptions, rather than the parent exception type. This is a separate issue from whether you have one big try or lots of little trys.
    NOTE HOWEVER THAT YOU SHOULD NOT CATCH RuntimeException OR ITS DESCENDANTS. RuntimeException indicates an error in your code. Catching it is usually pointless, because you can't recover from it, and trying to continue on will probably lead to corrupt results at best.

  • Catch exceptions out of catch block

    Hello,
    Here in my job we was using Log4J , but since some of our logs need complex things like replication for
    some databases analise log t see where we will send in and lots of emails too , so we are developing our own log framework, my question is , there is a way to caught exceptions not declared in a try catch block?
    Example: IndexOutOfBound some time are not cought in a try catch block, or some files sometimes came null and throws uncauht exceptions some times, there is a way to me listen to all execptions wich was throwed to the system and have a no catch block??
    Thank You
    Ricardo

    When an uncaught exception occurs, the JAVA runtime system determins the ThreadGroup of the current thread and calls "uncaughtException" on that ThreadGroup object. However, the java.lang.ThreadGroup object performs the printing of the stack trace, which you will normally see in such cases.
    If you want to modify this behaviour, you could write your own ThreadGroup class, which extends java.lang.ThreadGroup and provides a different implementation of uncaughtException. E.g. you could write a log message to your own log file or whatever you want.
    Now you must guarantee, that all threads in your system are childs of that customized ThreadGroup. I did this in my main class:
    public class Application implements Runnable {
        private String[] argv;
        public static void main(String[] argv) {
            this.argv = argv; // save this for use in the run method
            ThreadGroup tg = new MyThreadGroup();
            // start a thread in the new thread group and terminate the original main thread
            new Thread(tg, this, "main").start();
        public void run() {
            // perform all other tasks here
    public class MyThreadGroup extends ThreadGroup {
        public MyThreadGroup()  {
            super("MyThreadGroup");
        public void uncaughtException(Thread thread, Throwable exception) {
            // ignore death of thread
            if (exception instanceof ThreadDeath)
                return;
            // ---- perform your actions here ----
    }All other threads, which are created later, are also in the new thread group, because they inherit the group from their creator.

  • Handling multiple exceptions with a single catch block

    In the following code:
    try{
    catch (NumberFormatException a) {
    catch (UserDefinedException b) {
    The code for both catch blocks are identical. Is there no way I can combine these into one block?
    For example, could I not do:
    catch ( (NumberFormatException a) || (UserDefinedException b) ){
    or anything similar?
    I did think of:
    try{
    try{
    catch (NumberFormatException a) {
    throw new UserDefinedException("");
    catch (UserDefinedException b){
    but this just seems to be a waste of code. Any ideas?

    I would use the fundamental way to combine identical sections of code--put the code in a new method:
        try {
        catch (NumberFormatException a) {
          myExceptionHandlingMethod(a);
        catch (UserDefinedException b) {
          myExceptionHandlingMethod(b);
      private void myExceptionHandlingMethod(Throwable t) {
      }

  • Can't find class because of try catch block ?!

    Hello,
    I'm using the JNI to use a java library from my c++ code.
    JNI can't find the "Comverse10" class below, but when i put the try catch block in createMessage in comment, FindClass succeeds ?!
    Unfortunatly i need the code inside the try block ;-)
    I tried a few things, but none of them worked:
    - let createMessage throw the exception (public void createMessage throws ElementAlreadyExistsException ), so there isn't a try catch block in createMessage => result: the same
    - make a "wrapper" class Comverse that has a Comverse10 object as attribute and just calls the corresponding Comverse10.function. Result: Comvers could be found, but not constructed (NewObject failed).
    Can someone tell me what is going on ?!
    Thank you,
    Pieter.
    //Comverse10 class
    public class Comverse10 {
    MultimediaMessage message;
    /** Creates a new instance of Comverse10 */
    public Comverse10() {
    public void createMessage() {
    TextMediaElement text1 = new TextMediaElement("Pieter");
    text1.setColor(Color.blue);
    SimpleSlide slide1 = new SimpleSlide();
    //if i put this try catch block in comment, it works ?!
    try{
    slide1.add(text1);
    catch(com.comverse.mms.mmspade.api.ElementAlreadyExistsException e){}
    MessageContent content = new MessageContent();
    content.addSlide(slide1);
    this.message = new MultimediaMessage();
    message.setContent(content);
    message.setSubject("Mijn subjectje");
    for those of you who are intersted: here's my C++ code:
    //creation of JVM
    HRESULT Java::CreateJavaVMdll()
         HRESULT HRv = S_OK;
    char classpath[1024];
         jint res;
         if(blog)     this->oDebugLog->Printf("CreateJavaVMdll()");
         strcpy(classpath,"-Djava.class.path="); /*This tells jvm that it is getting the class path*/
         strcat(classpath,getenv("PATH"));
         strcat(classpath,";D:\\Projects\\RingRing\\MMSComposer;C:\\Progra~1\\j2sdk1~1.1_0\\lib");     //;C:\\Comverse\\MMS_SDK\\SDK\\lib\\mail.jar;C:\\Comverse\\MMS_SDK\\SDK\\lib\\activation.jar;C:\\Comverse\\MMS_SDK\\SDK\\lib\\mmspade.jar
         //------Set Options for virtual machine
         options[0].optionString = "-Djava.compiler=NONE"; //JIT compiler
         options[1].optionString = classpath;                                        //CLASSPATH
         //------Set argument structure components
         vm_args.options = options;
         vm_args.nOptions = 2;
         vm_args.ignoreUnrecognized = JNI_TRUE;
         vm_args.version = JNI_VERSION_1_4;
         /* Win32 version */
         HINSTANCE hVM = LoadLibrary("C:\\Program Files\\j2sdk1.4.1_01\\jre\\bin\\client\\jvm.dll");
         if (hVM == NULL){
              if(blog) oDebugLog->Printf("Can't load jvm.dll");
              return E_FAIL;
         if(blog) oDebugLog->Printf("jvm.dll loaded\n");
         LPFNDLLFUNC1 func = (LPFNDLLFUNC1)GetProcAddress(hVM, "JNI_CreateJavaVM");
         if(!func){
              if(blog)     oDebugLog->Printf("Can't get ProcAddress of JNI_CreateJavaVM");
              FreeLibrary(hVM);     hVM = NULL;
              return E_FAIL;
         if(blog)     oDebugLog->Printf("ProcAddress found");
         res = func(&jvm,(void**)&env,&vm_args);
         if (res < 0) {
    if(blog)     oDebugLog->Printf("Can't create JVM with JNI_CreateJavaVM %d\n",res);
    return E_FAIL;
         if(blog)     oDebugLog->Printf("JVM created");
         return HRv;
    //finding Comverse10 class:
    HRESULT CALLAS MMSComposer::InitializeJNI(void)
         HRESULT HRv=E_FAIL;
         DWORD T=0;
         try
              if(blog)     oDebugLog->Printf("\nInitializeJNI()");
              bJVM = FALSE;
              jni = new Java(oDebugLog);
              if(jni->CreateJavaVMdll()!=S_OK){
                   if(blog)     oDebugLog->Printf("CreateJavaVMdll() failed");     
                   return HRv;
              jclass jcls = jni->env->FindClass("Comverse10");
              if (jcls == 0) {
    if(blog)     oDebugLog->Printf("Can't find Comverse10 class");
                   jclass jcls2 = jni->env->FindClass("test");
                   if (jcls2 == 0) {
                        if(blog)     oDebugLog->Printf("Can't find test class");
                        return HRv;
                   if(blog)     oDebugLog->Printf("test class found %08x",jcls2);
    return HRv;
              if(blog)     oDebugLog->Printf("Comverse10 class found %08x",jcls);
              jmethodID mid = jni->env->GetMethodID(jcls , "<init>", "()V");
              if (mid == 0) {
                   if(blog)     oDebugLog->Printf("Can't find Comverse10() constructor");
    return HRv;
              if(blog)     oDebugLog->Printf("Comverse10() constructor found");
              jobject jobj = jni->env->NewObject(jcls,mid);
              if(jobj==0)
                   if(blog)     oDebugLog->Printf("Can't construct a Comverse10 object");
    return HRv;
              if(blog)     oDebugLog->Printf("Comverse10 object constucted");
              //Create Global reference, so java garbage collector won't delete it
              jni->jobj_comv = jni->env->NewGlobalRef(jobj);
              if(jni->jobj_comv==0)
                   if(blog)     oDebugLog->Printf("Can't create global reference to Comverse10 object");
    return HRv;
              if(blog)     oDebugLog->Printf("global reference to Comverse10 object %08x created",jni->jobj_comv);
              bJVM=TRUE;
              HRv=S_OK;
         }     catch( IDB * bgError ) { throw bgError->ErrorTrace("InitializeJNI::~InitializeJNI",HRv, 0, T); }
              catch(...) { throw IDB::NewErrorTrace("InitializeJNI::~InitializeJNI",HRv, 0, T ); }
              return HRv;

    >
    I would guess that the real problem is that that the
    exception you are catching is not in the class path
    that you are defining.Thanks jschell, that was indeed the case.
    I don't have the docs, but I would guess that
    FindClass() only returns null if an exception is
    thrown. And you are not checking for the exception.
    Which would tell you the problem.Ok, i'll remember that. But what with exceptions thrown in my java code, the documents say
    // jthrowable ExceptionOccurred(JNIEnv *env);
    // Determines if an exception is being thrown. The exception stays being thrown until either the native code calls ExceptionClear(), or the Java code handles the exception
    so, what if the java code throws an exception and catches it, will i be able to see that in my c++ code with ExceptionOccurred ?
    or
    should the java method be declared to throw the exception (and not catch it inside the method)
    Again, thank you for your help, it's greatly appreciated !

  • Exception Handling In BPEL  By using Catch Blocks or Fault Policies Or Both

    I have a confusion regarding
    Exception handling :
    When Should i go for 1)Catch Block (Remote , or binding ) in bpel for exception handling .
    2)Fault Policy , Fault binding.xml
    Currently iam using catch blocks , but even fault policy is good , but can i use both...
    Currently in My bpel ,when any error occurs i have to send a error notification by Email .
    Currently i have exposed the email service which shuts emails and write a file with errored Message.
    Hence if any error i will catch i in a parent BPEL, i will just invoke the above email, service .
    So anybody can help me by giving the suggestion how to go for the best approach
    Edited by: anantwag on Mar 23, 2011 6:31 AM

    Currently in My bpel ,when any error occurs i have to send a error notification by Email .
    Currently i have exposed the email service which shuts emails and write a file with errored Message.Seeing your use case I will suggest you to use fault handling framework (fault policy). Fault handling framework should be used where you need generic error handling framework which handles all the faults occured in any composite component. Generally BPEL catch block should be used to propagate error info/fault back to the client/to fault handling framework or to consume an error
    Regards,
    Anuj

  • Return in finally block hides exception in catch block

    Hi,
    if the line marked with "!!!" is commented out, the re-thrown exception is not seen by the caller (because of the return in the finally block). If the line is commented in, it works as I would expect. Is this is bug or a feature? I assume the return removes the call to troubleMaker() from the call stack - including the associated exception...
    Christian Treber
    [email protected]
    public class ExceptionDemo extends TestCase
    public void testException()
    String lResult = "[not set]";
    try
    lResult = troubleMaker();
    System.out.println("No exception from troubleMaker()");
    } catch(Exception e)
    System.out.println("Caught an exception from troubleMaker(): " + e);
    System.out.println("Result: " + lResult);
    public String troubleMaker()
    boolean lErrorP = false;
    try
    if(6 > 5)
    throw new RuntimeException("Initial exception");
    return "normal";
    } catch (RuntimeException e)
    System.out.println("Caught runtime exception " + e);
    lErrorP = true;
    throw new RuntimeException("Exception within catch");
    } finally
    System.out.println("Finally! Error: " + lErrorP);
    if(!lErrorP)
    return "finally";

    This is specified in the Java Language Specification, section 14.19.2 .
    -- quote
    If execution of the try block completes abruptly because of a throw of a value V, then there is a choice:
    * If the run-time type of V is assignable to the parameter of any catch clause of the try statement, then the first (leftmost) such catch clause is selected. The value V is assigned to the parameter of the selected catch clause, and the Block of that catch clause is executed. Then there is a choice:
    o If the catch block completes normally, then the finally block is executed. Then there is a choice:
    + If the finally block completes abruptly for any reason, then the try statement completes abruptly for the same reason.
    -- end quote
    "completing abruptly" (in this instance) means a throw or a return, as specified in section 14.1
    Ok? It's not a bug, it's the defined behaviour of the language. And when you think about it, this behaviour gives you the option to do what you want in your finally block.

  • Is it possible to get back to the Try block from a Catch block?

    I would like to perform validations on a config file. To do so, I have a method which will throw a MissingConfigException or an InvalidConfigException if something is missing or if something doesn't match a regex pattern. Although I have to do that with many different configs and I would like to go through all of them so that my error message will report all missing / incorrect configs at once instead of 1 at a time. Thus, I wanted to know if there would be any way I could do something like:
    try {
         config.validate("ServerPort","\\d+");
         config.validate("MaxClients","\\d+");
    } catch(MissingConfigException e) {
    } catch(InvalidConfigException e) {
    and have the catch block executed before going through the try block from where the exception was thrown.
    Thanks in advance!                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                               

    No, you can't jump back into the try. You can put your try/catch inside a loop though, so that if one fails, it will loop around and start a new try for the next one. Define a class that holds all the data you need to pass to validate, and make an array or List of those objects. Change or add a validate signature that takes this object, or just extract the strings from each one when you call validate.
    ThingToValidate[] val = new ThingToValidate[] { new ThingToValidate("ServerPort", "\\d+"),
        new ThingToValidate("MaxClients", "\\d+"),
    for (int ix = 0; ix < val.length; ix++) {
        try {
            config.validate(val[ix]);
            // OR
            config.validate(val[ix].getThingy, val[ix].getOtherThingy[]);
        catch (....) {
            // logger.warn("failed...");

  • TRY..catch block in a loop.

    Hi Epxerts,
    Can i use loop endloop for the following TRY..CATCH block as i need to send multiple records
    LOOP.
    TRY.
      Assign row
        it_prxstruc-MT_table_EXTRACT-record-row = wa_area1. --> one record of
        CALL METHOD prxy->execute_asynchronous
          EXPORTING
            output = it_prxstruc --> 1 record.
        COMMIT WORK
      CATCH cx_ai_system_fault .
        DATA fault TYPE REF TO cx_ai_system_fault .
        CREATE OBJECT fault.
        WRITE :/ fault->errortext.
    ENDTRY.
    ENDLOOP.
    Please suggest accordingly.
    Thanks
    Dany

    Try-Endtry specifies the territory of exception that catch is going to handle. Do it this way..
    TRY.
    LOOP.
    Assign row
    it_prxstruc-MT_table_EXTRACT-record-row = wa_area1. --> one record of
    CALL METHOD prxy->execute_asynchronous
    EXPORTING
    output = it_prxstruc --> 1 record.
    COMMIT WORK
    ENDLOOP.
    CATCH cx_ai_system_fault .
    DATA fault TYPE REF TO cx_ai_system_fault .
    CREATE OBJECT fault.
    WRITE :/ fault->errortext.
    ENDTRY.
    G@urav.

  • RAISERROR with Try/Catch does not exit after exception in catch block

    I am trying to propogate an error from within my proc out to the caller.
    In the attached example I have 2 sets of try catch blocks.
    I raiserror in the first
    catch the error and then raiserror again. (I expect to exit)
    I do not expect :
    to get to print 'post test'
    to get to second try block.
    but this does not exit, instead the code flows as per 2 runs.
    I do not understand the reason for the flows, as it seems counterintuitive to be raising an error but then still print following exceptions. I cannot seem to find any references that explains this behaviour.
     running tests together results
    print '-------------------------------------------------------'
    print 'test 15'
    exec test_raiseerror 15
    print '-------------------------------------------------------'
    print 'test 16'
    exec test_raiseerror 16
    print '-------------------------------------------------------'
    print 'test 17'
    exec test_raiseerror 17
    print '-------------------------------------------------------'
    print 'test 18'
    exec test_raiseerror 18
    print '-------------------------------------------------------'
    'RESULTS'
    test 15
    error number provided: 15
    Msg 50000, Level 15, State 1, Procedure test_raiseerror, Line 21
    name hello 15
    post test
    15
    Msg 50000, Level 15, State 1, Procedure test_raiseerror, Line 37
    name hello 2 15
    post test2
    test 16
    error number provided: 16
    Msg 50000, Level 16, State 1, Procedure test_raiseerror, Line 21
    name hello 16
    post test
    16
    Msg 50000, Level 16, State 1, Procedure test_raiseerror, Line 37
    name hello 2 16
    post test2
    test 17
    error number provided: 17
    post test
    17
    post test2
    test 18
    error number provided: 18
    post test
    18
    post test2
    Msg 50000, Level 17, State 1, Procedure test_raiseerror, Line 21
    name hello 17
    Msg 50000, Level 17, State 1, Procedure test_raiseerror, Line 37
    name hello 2 17
    Msg 50000, Level 18, State 1, Procedure test_raiseerror, Line 21
    name hello 18
    Msg 50000, Level 18, State 1, Procedure test_raiseerror, Line 37
    name hello 2 18
    run tests seperately
    exec test_raiseerror 15
    error number provided: 15
    RESULTS 15
    Msg 50000, Level 15, State 1, Procedure test_raiseerror, Line 21
    name hello 15
    post test
    15
    Msg 50000, Level 15, State 1, Procedure test_raiseerror, Line 37
    name hello 2 15
    post test2
    exec test_raiseerror 16
    RESULTS 16
    error number provided: 16
    Msg 50000, Level 16, State 1, Procedure test_raiseerror, Line 21
    name hello 16
    post test
    16
    Msg 50000, Level 16, State 1, Procedure test_raiseerror, Line 37
    name hello 2 16
    post test2
    exec test_raiseerror 17
    RESULTS 17
    error number provided: 17
    post test
    17
    post test2
    Msg 50000, Level 17, State 1, Procedure test_raiseerror, Line 21
    name hello 17
    Msg 50000, Level 17, State 1, Procedure test_raiseerror, Line 37
    name hello 2 17
    exec test_raiseerror 18
    RESULTS 18
    error number provided: 18
    post test
    18
    post test2
    Msg 50000, Level 18, State 1, Procedure test_raiseerror, Line 21
    name hello 18
    Msg 50000, Level 18, State 1, Procedure test_raiseerror, Line 37
    name hello 2 18
     CODEBLOCK:
    if object_id('test_raiseerror','P') is not null
    drop proc test_raiseerror
    go
    create proc test_raiseerror(@id as int) as
    begin
    begin try
    declare @name varchar(20)
    select @name = 'hello'
    raiserror('name %s %d',@id,1,@name,@id)
    print 'next'
    end try
    begin catch
    declare @errormessage nvarchar(4000)
    declare @errornum int
    select @errormessage = error_message()
    , @errornum = error_severity()
    print 'error number provided: ' + convert(varchar(2),@errornum)
    raiserror(@errormessage, @errornum,1)
    print 'post test'
    end catch
    begin try
    select @name = 'hello 2'
    raiserror('name %s %d', @id,1,@name, @id)
    end try
    begin catch
    select @errormessage = error_message()
    , @errornum = error_severity()
    print @errornum
    raiserror(@errormessage, @errornum,1)
    print 'post test2'
    end catch
    end
    go
    sqlserver 2008 & 2008 R2

    There is a Connect that describes a similiar complaint.  But basically a raiserror inside a catch block does not terminate the procedure, it will continue with any additional code in the CATCH and FINALLY unless it hits a return statement.
    http://connect.microsoft.com/SQLServer/feedback/details/275308/have-raiserror-work-with-xact-abort

Maybe you are looking for

  • Error in the registries, 4-20mA

    I´m using fieldpoints (FP-2020) to register the values (4-20mA) originating of a CFI (converter frequency - current). In some occasions fieldpoints have registered erroneos values, as if the CFI was not connected. Is it possible that fieldpoint varie

  • How to get old search box back

    Yesterday my broswer updated and I now have version 36. This comes with a new search box. I had several search engines listed with different languages, now all i see in this stupid new version is the same icon, I can't see the labels anymore which ha

  • Lightroom 3.2 external editor problem

    Hello, I have been experiencing problems with using external editors in Lightroom 3.2.  Currently there are 3 external editors which display a problem (but all of which work fine as an external editor or stand-alone program).  Those are Alien Skin's

  • Use of Tab key on parameter form

    Hi, On Reports 9 parameter form the tab key is not working. Is there any way to enable tab key? Thanks, Dean

  • Reference Field for Document Type AB has made Mandatory

    Dear Experts,     In OBA7, for Document Type AB (Accounting Document) I had clicked the Reference Number Field Check Box (Made it mandatory)     So it became mandatory every time to enter the Reference Number when using the Document Type AB     Now I