On 17/07/2013 12:43, Mark Thomas wrote:
> Violeta Georgieva <miles...@gmail.com> wrote:
>
>> Having the example from the test:
>> (x->y->x-y)(2)(1)
>>
>> We may transform it like this:
>> (()->y->2-y)()(1)
>>
>> So instead of returning right way when formalParameters.isEmpty()
>>
>> Shouldn't we have something like the one below in order to continue
>> evaluation of the nested expression:
>>
>>
>>
>> --- org/apache/el/parser/AstLambdaExpression.java (revision 1504024)
>> +++ org/apache/el/parser/AstLambdaExpression.java (working copy)
>> @@ -54,9 +54,12 @@
>> LambdaExpression le = new LambdaExpression(formalParameters, ve);
>> le.setELContext(ctx);
>>
>> + Object result = le;
>> + int i = 2;
>> if (formalParameters.isEmpty()) {
>> // No formal parameters - invoke the expression
>> - return le.invoke(ctx, (Object[]) null);
>> + result = ((LambdaExpression) result).invoke(ctx,
>> (Object[])
>> null);
>> + i++;
>> }
>>
>> // If there are method parameters, need to invoke the expression
>> with
>> @@ -64,8 +67,6 @@
>> // should be that many nested expressions.
>> // If there are more nested expressions that parameters this will
>> return
>> // a LambdaExpression
>> - Object result = le;
>> - int i = 2;
>> while (result instanceof LambdaExpression && i <
>> jjtGetNumChildren()) {
>> result = ((LambdaExpression) result).invoke(
>> ((AstMethodParameters)
>> children[i]).getParameters(ctx));
>
> I agree the current solution isn't right but I have some local test cases
> that I'm fairly sure will still fail with the above patch. I'm working on a
> fix and should have something to commit soon.
Fix applied. What do you think?
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
