Notes for TAG f2f: Error handling

Henry S. Thompson
5 March 2009

1. Error handling

My recent contribution to the "HTML and XML" thread ( wasn't actually about errors.

However it does come back to errors at some points. I find Sneddon's 0060 reasonably measured, and asking for some do-able things (Hah: 3023bis raises its head again)

I guess we should forward the concerns about serialisation, particularly PHP but as a general matter, to the XML Core WG. It's arguably the case that they should consider producing an Infoset Serialization spec. See Sneddon's 0063.

Noah's 0181 is a good summary of the language spec./behaviour spec. distinction and why it's right.

Although I think he gets a bit carried away at other points, I like ERH's 0129

1.1. What needs to be taken forward and how?

  1. Prevalence of XML serializers which produce non-WF output: which is the right forum/mechanisms: Talk to PHP implementors? [XML Core WG?]
  2. XML Spec's (lack of) definition of error handling
    1. Reach consensus about the facts of the matter [HST to prepare]
    2. Decide what if anything to do [TAG discussion]
  3. News from the front lines: Opera has joined XML Core WG, sent Simon Pieters with a draft rewrite of the XML Stylesheet PI REC with considerable elaboration in the error handling area. . .
  4. Declarative specifications of error recovery: Tagsoup, PyXUP, . . . [Outside research activities]
  5. Language definition vs. consumer behaviour specification: What should be in a specification? [TAG discussion?]
  6. Scope of possible actions in this area: Who else is doing or willing do do some of this? How can we actually get relevant constituencies to move? [TAG metadiscussion]