attribute resolver script

[Peter Schober]

* Peter Schober <peter.schober at univie.ac.at> [2016-11-09 15:44]:
> * Michael Dahlberg <olgamirth at gmail.com> [2016-11-09 15:34]:
> > Any suggestions on why/when you would choose a mapped attribute over
> > a scripted attribute?
> 
> I'd chose any other method over scripted, if there are alternatives.
> 
> * configuration > programming
>   (though maybe Scott's comments about Mapped specifically should be
>   taken to mean there might be exceptions)
>   
> * no issues wrt Java and scripting engine version (Rhino vs. Nashorn)
>   (or adding and updating scripting engines for other languagues,
>   e.g. JPython)
> 
> * performance, at least theoretically

Oh, and there's:

* Chances of messing it up.
There's an almost infinitely higher chance for errors in custom code
compared to using one of the "curated" methods such as Template or
Mapped.

* Error reporting.
Often a script will fail without any indication what specifically went
wrong within the script. In contrast errors in applying the
appropriate configuration should be caught more easily and have the
potential (at least) for much easier to understand error messages.
-peter