Meeting minutes
Accept the minutes of the previous meeting
We think they're OK
Review of open actions
[No news]
Status reports
Norm: Bug fixes
Steven: I'm writing my MarkupUK paper
Norm: I updated the references
Serialization
Steven: We just say "serialization", without detail
Norm: Slippery slope
Steven: Leave as is?
David: Perhaps, say that serialization is implementation dependent
Norm: "Implementation defined"
John: Serialization is actually translation from parse tree to XML output tree, and serialization is converting that to text, so I think "serialization" is the wrong term.
David: "Serialization" is about ordering.
… it might be helpful to use another term.
John: So then entities is not part of ixml
Norm: The spec requires serialized characters
Bethan: I'm not convinced that the spec's "serialization" doesn't really mean serialization.
John: I don't serialize to characters, but to a model
Steven: I think the ixml spec (should) allow serializing to memory
Bethan: I think the current spec intends an understanding of XML serialization
Steven: I certainly didn't write it that way, and we don't normatively reference it
David: Calling it serialization is more than mildly confusing
Norm: I think we should take the character-wise interpretation
John: We've open a small can of worms. It's worth thinking about over a longer period.
Pragmas
Bethan: There has been a lot of email discussion
Norm: Item 9
Steven: I think that the content of pragmas should be able to do that, but not necessary encoding that in the spec.
Bethan: Should the spec say what it means when a pragma is in a particular position in the grammar
<Steven> s/Should/
Norm: The position of a pragma should have a predictable place in the serialization
Bethan: Pragmas have a syntactical role, but not a semantic role
John: It provides a well-defined place where it is, which may affect scope
… it is defined where it is in the tree, and the semantics of that pragma defines the scope
Bethan: I think we agree
David: Item 9 doesn't specify that
… I could have a pragma at the beginning, that annotates something deep in the grammar.
Bethan: I think 9 is a question of attachment and not annotation
… to which constructs in the grammar should it be possible to attach a pragma?
… 9 and 10 should both say attach and not annotate
Norm: What do we mean by 'attach'?
Bethan: Attachment is syntactic, and annotate is semantic.
John: Suppose you had a pragma setting a priority; you could put it anywhere in the rule; it depends on the implementation.
… there are others where the pragma has to be very close to the thing it is talking about
Bethan: Not sure if that is true. You could put them all at the root, it wouldn't matter.
… I don't see any example where the pragma has to be absolutely attached to anything
Steven: Is 'attached' just placement in the grammar?
Bethan: Yes.
Steven: I still don't understand what 'attach' means
David: Nowhere do we say that annotation and attachment are two sides of the operation
Bethan: It may be best to talk about syntactic scope
Norm: I propose that "attach" means 'is a child of'
<norm> in the XML representation of the grammar
Steven: I think then I accept 9 and 10
We agree on a version of 11 without 'minimal', and agree on 12