(moved from SyntaxConsiderations)
[SeanMcGrath RefactorOk] Please, please avoid going down the Infoset road. Yes, use XML syntax, just enough XML to get the job done yet still be easily "round-trippable" and "lossless."
I agree with Joe [AsbjornUlsberg]
[JeremyGray] 0 to infoset. +1 to namespaces. +1 to core in a single namespace. +1 to core namespace as default (saves bandwidth, increases ViewSourceClan clarity, and is processed no differently by real XML tools). -1 to modules in core namespace (should be in a separate namespace, and I would much prefer one per module.). The core namespace should be the baseline of what is required and what is optional but generally expected to work in Atom-powered applications.
[JamesSnell] Ok, well, It appears I'm not going to get my way so I'll concede the point of not having an infoset model... BUT, with that being the case, we need to be extra careful to fully articulate our model while describing the syntax.
[JoeGregorio] Please, no infosets. Just straight XML syntax. As for namespaces, I would like to see the 'core' of this new format be in a single namespace. The core needs to be simple to 'view-source' and implement. Using a slew of different elements from different namespaces would make that more difficult. I wouldn't mind the single namespace chosen being the nil namespace, though I am sympathetic to arguments against that. I would also like to see as many of the optional modules in the core namespace as well.
[DannyAyers] -1 to infoset (I'd favour an RDF/OWL version, but that doesn't look like happening), +1 to namespaces. +1 to core in a single namespace, +1 to we need to be extra careful to fully articulate our model while describing the syntax.
[RahulDave] On Infoset vs XML syntax, why not define the Infoset in XML syntax. Using DTD?XML Schema, even. I mean for validation, rather than for documentation. Once a precise documentation of one representation can be achived, one can come out of the woodwork with different equivalent carry's: Echo in RSS3, in SOAP, in RFC822, whatever is appropriate to an apps consumption and production. Obviously, the Echo XML syntax ought to be the meeting point for interop. (copied from NamespaceDiscussion)
I think it is a good idea to have an Infoset; after all, XML tagged text may not be the only serialization. There have been suggestions on this wiki to use yaml, etc.
[BillDehora]. An Infoset is a step backwards. Better to define the transform from one syntax to another. A non-XML Infoset will requires codecs to generate the events, codecs historically don't interoperate without tight coupling. Which is one reason XML became so useful for shuttling data about in the first place...
[DanielCazzulino] Infoset a step backwards?? Infoset IS the abstract way of defining functionality and data structures. XML is just a serialization format for the corresponding infoset. If you introduce a good enough Infoset, then the syntax may be switched, but the core remains. MS is pushing hard towards all-infoset XML implementations in the next version of .NET. We should struggle to do the same. I think many people doesn't know what an infoset is, that's why they don't want it. I suggest investigation. It's the core technology for all W3C XML-related specs. Why would we depart from that?