Creating a new setChild method should be fine. How would we differentiate between + and +=? Two new methods? Or something else?
On Nov 16, 2015, at 9:14 PM, Alex Harui <aha...@adobe.com> wrote: > > > On 11/16/15, 10:54 AM, "Harbs" <harbs.li...@gmail.com> wrote: > >> >> On Nov 16, 2015, at 8:03 PM, Alex Harui <aha...@adobe.com> wrote: >> >>> >>> >>> On 11/16/15, 9:54 AM, "Harbs" <harbs.li...@gmail.com> wrote: >>> >>>>> 2) detect read and write on XML/XMLList and convert read to child(). >>>>> What >>>>> about write()? >>>> >>>> Good question. I think we should map xml.foo = <foo id=“1"/> to >>>> xml.replace(“foo”,new XML(“<foo id=\”1\”/>”)); >>> >>> What does the above actually do if there is only one foo child and if >>> there are more than one foo.child? >> >> 1. xml.foo = someXML replaces all foos with someXML (in the location of >> the first foo) >> 2. xml.foo = someXMLList removes all foos and adds in the ones from the >> list (in the location of the first foo) >> >>> Do you know if you can use += to append? >> >> How do you like that? I’ve never thought of trying +=! I just tried and >> it works like a charm! It adds it in after the last foo or at the end if >> foo does not exist. > > Hmm. I wonder what other operators work. Like “-=“. And does xmlList1 + > xmlList2 concatenate them? > >> >> I’m not sure what the best mapping for this will be. We probably should >> get a reference to the last foo and map it to >> xml.insertChildAfter(lastFoo,newFoo) if there are any foos. Otherwise it >> should map to appendChild(). > > OK. Thanks for doing the testing. > > The compiler cannot know whether there will be a foo at compile-time, so > for all of these cases, I think we’ll have to call a new setChild() method > and your code will have to “do the right things”. > > Replacing “=“ expressions with method calls is quite painful in the > compiler and the output can look very messy. What do you think about > implementing setChild as a setter instead of a method? > > -Alex
