[Laszlo-dev] Is this compiled code for a method, in debug, mode, correct?

[P T Withington]

To elaborate a little more on André's comment:

$lzsc$thrownError is an error that was thrown from a block where  
`#pragma "throwsErrors=true" was used to _prevent_ reporting (and  
catching) of the error.  (I refer to this as "guarding" the error  
below.)

So, if the error we catch matches that "guarded" error, we _don't_  
report it, we just immediately re-throw it.  Any other error would  
mean that the error is _not_ guarded (either it is the original error,  
or it superseded a guarded error), so that error _will_ be reported by  
the debugger (and then fall through and be re-thrown).

There is a slight inefficiency here.  The debugger uses a separate  
memo to avoid re-reporting unguarded errors, but we could optimize  
that by, once the debugger has reported the error, then setting $lzsc 
$thrownError to the reported error, to avoid recalling the exception  
code as the error percolates up the stack.

On 2009-09-21, at 15:11, André Bargull wrote:

> The error is only discarded if "catcherrors" is enabled (and the  
> "throwsError" pragma isn't used).
>
>> I was looking at the as3 generated code for an onclick handler  
>> which came
>> from this source code
>>
>>       <button onclick="canvas.test(1)">click me </button>
>>
>> It looks like this, when compiled in debug mode
>>
>> public function $lzc$handle_onclick$$args$2Elzx_11_41_$m492 ($lzc 
>> $ignore_$1)
>> {
>>    try {
>>        canvas.test(1);
>>    }
>>    catch ($lzsc$e) {
>>        if ($lzsc$e is Error) {
>>            if (lz["$lzsc$thrownError"] === $lzsc$e) {
>>                throw $lzsc$e;
>>            };
>>            $lzsc$runtime.reportException("args.lzx", 11, $lzsc$e);
>>        };
>>        throw $lzsc$e;
>>    }
>> }
>>
>> If some error is thrown in the call to "test", this wrapper looks  
>> like it
>> will always catch and then
>> rethrow it. Is this the expected behavior? I sort of thought we  
>> were going
>> to catch and report the error
>> to our debugger, and then discard it rather than rethrowing it, in  
>> this case
>> (debugger on)?
>>
>> -- 
>> Henry Minsky
>> Software Architect
>> hminsky at laszlosystems.com <http://www.openlaszlo.org/mailman/listinfo/laszlo-dev 
>> >
>>
>