bug in exception handling (begin try... begin catch...)?

begin try
-- insert ...
select * from nonexistingtableorfunction;
end try
begin catch
print 'catch me if you can'; -- line will not be printed
end catch;
begin try
exec doNotExitingprocedure;
end try
begin catch
print 'this line is printed :-)';
end catch;
Why this different behaviour?
If you wrap 1. example in anthother proc with exception handling
(begin try... begin catch...),
the wrapping proc will catch the error.
The above behaviour is not what I would expect compared with other
languages such as Java, C#, PL/SQL...
Greetings
B. D. JensenI think this is one of numerous reasons I have noticed a distinct lack of
uptake on the new TRY mechanism. Until it really does catch errors I won't
use it, nor advise my clients to.
"B D Jensen" <bjorn.d.jensen@.gmail.com> wrote in message
news:1191585968.457460.319780@.o80g2000hse.googlegroups.com...
> begin try
> -- insert ...
> select * from nonexistingtableorfunction;
> end try
> begin catch
> print 'catch me if you can'; -- line will not be printed
> end catch;
>
> begin try
> exec doNotExitingprocedure;
> end try
> begin catch
> print 'this line is printed :-)';
> end catch;
>
> Why this different behaviour?
> If you wrap 1. example in anthother proc with exception handling
> (begin try... begin catch...),
> the wrapping proc will catch the error.
>
> The above behaviour is not what I would expect compared with other
> languages such as Java, C#, PL/SQL...
> Greetings
> B. D. Jensen
>|||B D,
I believe that what is happening here is that in your first example, the
batch is failing with a syntax error:
Msg 208, Level 16, State 1, Line 1
Invalid object name 'nonexistingtableorfunction'.
This means that it failed at compile time and never executed. Since it
never executed, the try / catch had no chance to do anything.
In your second example, the called routine got the same error as above and
never executed.
However, in this case the calling routine is running just fine and the try /
catch sees that the called routine got an error.
At least, that is what it looks like to me.
RLF
"B D Jensen" <bjorn.d.jensen@.gmail.com> wrote in message
news:1191585968.457460.319780@.o80g2000hse.googlegroups.com...
> begin try
> -- insert ...
> select * from nonexistingtableorfunction;
> end try
> begin catch
> print 'catch me if you can'; -- line will not be printed
> end catch;
>
> begin try
> exec doNotExitingprocedure;
> end try
> begin catch
> print 'this line is printed :-)';
> end catch;
>
> Why this different behaviour?
> If you wrap 1. example in anthother proc with exception handling
> (begin try... begin catch...),
> the wrapping proc will catch the error.
>
> The above behaviour is not what I would expect compared with other
> languages such as Java, C#, PL/SQL...
> Greetings
> B. D. Jensen
>|||Take a look at the BOL here:
ms-help://MS.SQLCC.v9/MS.SQLSVR.v9.en/udb9/html/3a5711f5-4f6d-49e8-b1eb-53645181bc40.htm
In your first case, the missing object caused the batch to terminate, and
TRY/CATCH won't help if the batch is terminated. In your second case, the
batch is not terminated and TRY/CATCH works.
Yuu can place a PRINT statement at the end of each batch to see whether the
batch is terminated before it reaches the end.
Linchi
"B D Jensen" wrote:
> begin try
> -- insert ...
> select * from nonexistingtableorfunction;
> end try
> begin catch
> print 'catch me if you can'; -- line will not be printed
> end catch;
>
> begin try
> exec doNotExitingprocedure;
> end try
> begin catch
> print 'this line is printed :-)';
> end catch;
>
> Why this different behaviour?
> If you wrap 1. example in anthother proc with exception handling
> (begin try... begin catch...),
> the wrapping proc will catch the error.
>
> The above behaviour is not what I would expect compared with other
> languages such as Java, C#, PL/SQL...
> Greetings
> B. D. Jensen
>