Mule 3.7 / 3.8中的Java组件与全局函数

Question

For some of our flows, we'll have some output attributes (a checksum for instance) that we can't generate directly through DataWeave so we chose to calculate them using Java functions and we have too possibilities :

• Use a Java component and put the result on a variable using a transformer class
• Define a global function that calls a static method from a class and use it in a Transform Message component

I know that we can have some performance problems with static method as they can't be garbage collected. What are the pros and cons of these 2 choices ?

Answer 1

I think it depends on the complexity of the function. Anything you can do in Java you can do directly using MVEL in a global function - no need to call out to a Java static method.

The main benefit of using a global function is that you can use MEL/MVEL that may make the function a lot less verbose because of handy MVEL features like property navigation, folds/projections etc. and allows you to access Mule vars/props/payload easily using MEL like #[flowVars.].

But I think if it's a very complicated function then a Java component may be easier for readability/maintenance etc. Also a Java component may be easier to modularise and share with other projects separately.

There may be performance benefits with one over the other, but potentially negligible or you could profile it to see the performance comparison.