RE: First draft of wiki text for 'scala-xml' project in Scala Incubator

Anthony B. Coates wrote:

> Comments, etc. would be very welcome.

It's a great list of questions - what's the Scala community way of responding? I can list post all the questions with inline comments, or one post per question, but maybe that isn't going to be the most convenient format in which to have a discussion with more than a couple of participants?

In general, I think the answers should depend a lot on whether in each case we think we can come up with an implementation that does what a working approximation of everyone would expect a particular feature to do. For example, I would really hope that we could come up with an efficient W3C-compliant document representation so that it is rarely necessary to roll a new one or import a Java one. On the other hand, I'd be amazed if we could come up with the ideal huge document management scheme, because it's still a thorny research issue in the XML community in general.

ISTM that we also need to look at some Scala-specific questions, such as

- How cute does the new setup look compared to the old one? Cuteness seems to me to be the headline feature of the current system, ie no-fuss XML literals and XML processing. At the risk of committing a cultural faux pas, the Perl dogma of making the easy things easy and the hard things possible seems a good fit here. For example, support for nested namespace bindings should add as little complexity as possible to the treatment of documents that are entirely in the null namespace, or which are entirely in one particular namespace. I'm expecting the "cute syntax" test to be the most important one for many existing Scala users, especially the ones who don't like XML but have to use it from time to time.

- How does XML [im]?mutability work? The Scala line in the general case seems to be that there are times when both are useful, with a preference for immutability. The argument has been made on this list that there are good reason for XML trees to be immutable. I'd expect the wider Scala community to be interested in that aspect of any new XML technology.

- How easy it will be to build on top of whatever core technology is implemented? In other words, if we implement infoset support, will it be possible for third parties to grow that to PSVI or XDM support in a way that doesn't look like a kludge? If we can make sure that the answer to this sort of question is "yes", we can get away with doing less of the job as The One Scala XML Way, and that seems like A Good Thing to me. If the answer is "no", how useful is infoset support that can't be used to build things that are based on infosets?