The ?. in Tapestry works for me (we use it a bit). Best of both worlds then.
mrg On Wed, Oct 9, 2013 at 3:02 PM, Andrus Adamchik <and...@objectstyle.org>wrote: > I am totally familiar with the issue. BTW here is the current Tapestry > approach to optional relationships (which IIRC evolved after trying > different things over the years) [1] : > > user.name // NPE is thrown > user?.name // Null is returned > > Andrus > > [1] http://tapestry.apache.org/property-expressions.html > > > > On Oct 9, 2013, at 2:29 PM, John Huss <johnth...@gmail.com> wrote: > > > Currently if you call PropertyUtils.getProperty with a chain of > properties, > > like "painting.name" and a part of the chain is null you get > > UnresolvablePathException("Null value in the middle of the path…"). I > > believe it was done this way to match > > org.apache.commons.beanutils.BeanUtils.getNestedProperty > > > > To me this is undesirable - I'd rather have it just return null and stop > > evaluation. That's how key value coding works in WebObjects. I just > have > > trouble seeing where an exception would be preferable to returning null. > > I'd be interested in either changing the behavior or adding an optional > > boolean flag that controls whether an exception is thrown or not. On the > > other hand, it's simple enough to create my own static method that wraps > > this call and catches the exception. > > > > Does anyone else have an opinion on this? > > > > John > >
