Minutes ixml call 2020-09-24

Agenda
Previous Actions
Namespace
Specification
Implementations
Test Suite
AOB
Next call

ACTIONS
ACTION: Steven to specify what happens when a name isn't an XML name
ACTION: Tom to look for an IXML domain.
ACTION: Steven to upload the spec to github
ACTION: Steven to research where to put S for attributes.

Next call
22 Oct 2020 14:00Z

Present: M, S, T

Previous Actions
ACTION: Steven to forbid "." at end of name
[withdrawn]
ACTION: Steven to specify what happens when a name isn't an XML name
[Ongoing]
ACTION: CMSMQ to send augmented grammar
[Done]
ACTION: Steven create Github repo
[Done: https://github.com/spemberton/ixml]

Namespace
T: We need a namespace URI
S: Whose namespace shall we use?
T: Good question. How about an ixml domain name?
ACTION: Tom to look for an IXML domain.
S: .org would be good.

Specification
ACTION: Steven to upload the spec to github
M: Whitespace and terminal retention in the resultant XML.
S: WHat should be returned int he XML for ism?
M: Yes.
T: I'm not worried about WH, just the terminals.
M: I'm not sure either way.
S: I don't care either way; if we can find justification for either, we 
should go with it.
T: Even though I started this discussion, it is not a hill I want to
die on. Leaving the terminals out makes prettyprinting easier. But
they aren't meaningful.
S: I can put both in the spec, and we can make the decision later.
M: Two arguments: 1) round tripping is easier if we preserve the
layout 2) there are some places in terminals where they are not marked
as excluded.
T: Example?
M: Whitespace between members of a character set.
S: The rule for S starts -S. All things in S except for comments are 
suppressed.
M: OK. If there were a comment in a range, you couldn't decide where to put 
it on roundtripping.
T: We shouldn't guarantee round tripping {something about isotopic}
M: We're only talking about IXML grammars.
T: I suppose it would be nice to make them roundtrippable.
M: We've already lost the whitespace.
M: But we have ["a"{comment}-"f"]
S: OK, I'll need to research where to put S for attributes.
ACTION: Steven to research where to put S for attributes.
S: Good catch.
M: Related question - is there a rule n: a1, c1, a2, c2, a3. that says 
attributes are handled first?
S: Yes. It's in the first rule for serialisation that attributes are
done before content; but it doesn't say that attributes are done in
order.
M: In XDM attributes are ordered.
T: I'm fascinated by the idea of reserializing.
S: It is still work to be done. I sketch a method in one of the papers, but 
I haven't done the work yet.

Implementations
S: You're giving a talk about yours in 2 weeks time.
T: Indeed. I think the rewrite is going to go well :-) [Description of 
work]

Test Suite
T: We exchanged some emails. I would like a testsuite that we can
generate from some self-describing document. Gherkin may not be the
right answer. xspec maybe. Later deciding on an XML format.
M: Later a non-XML format.
S: Use ixml.
M: I'm going to generate xpec cases; once that's going, I'll share. No 
progress this month.
S: It sounds like we each need to experiment, and then compare and merge.
T: Good plan.

AOB
[none]

#end