Meeting minutes
Previous Actions
Action: Steven to research where to put S for attributes.
<trackbot> Sorry, but no Tracker is associated with this channel.
Constinues
Action: Michael to comment on conformance section of new draft
<trackbot> Sorry, but no Tracker is associated with this channel.
Done
Action: Steven to apply Michael's text to spec
<trackbot> Sorry, but no Tracker is associated with this channel.
[Done]
Steven: Not yet pushed to github
https://
Steven: Please check.
https://
Steven: I don't think we answered those questions; not added to the spec.
ā¦ two questions unanswered there: code points, and grammars
MSM: I am confident that the spec should have a story if input doesn't obey the rules, it should be clear if it should be flagged as an error, or be 'undefined'.
<join_subline> š from toronto canada
MSM: if it must 'raise an error', then it protects us from future changes to the spec.
ā¦ I asked what SQL does, they said behaviour is undefined.
<join_subline> found this channel in https://
MSM: which allows processors to add their own features.
Steven: I'm a big fan of letting the user know (i.e. error)
MSM: I'm ok with that.
MSM: hex values MUST indicate an assignable unicode character.
Steven: And that is to protect output from nonsense Unicode sequences
Action: MSM to check conformance text in spec
Steven: All other actions continue
Namespaces/Text injection
Steven: Thanks for the new suggestions. I suggest that we create requirements for the next version, just to get a first version out of the door.
MSM: If we provide for declaration of namespaces, then namespace qualified elements and attributes just use a QName
ā¦ and then use a pragma/structured comment for the actual binding
ā¦ not local bindings
ā¦ NS injection may be easy.
ā¦ For text injection, the grammar has the same declarative meaning.
ā¦ it's the mirror image of the minus marker
ā¦ which gives it a certain appeal
John: And that could happen in any rule.
MSM: "Is it possible to create a nondeterministic grammar?"
Steven: Inserted text has no effect on the parse, only on the serialisation
Action: MSM to check conformance text in spec
<trackbot> Sorry, but no Tracker is associated with this channel.
<cmsmcq> https://
https://
Tom: I like suggestion 3.2
MSM: The aim is to target to a particular vocab, except it requires more. Getting <ol> is easy, but <list type="ordered"> is not
Tom: I would like plusses on nonterminals as well as terminal text.
MSM: We don't need it. There is a way to do that.
Tom: If we expect users to understand "+", we should make it do the same thing.
Steven: We dont need plus on a nonterminal, because they are already plussed.
MSM: It is like the ^ symbol.
Tom: Plus feels right.
Steven: We have the QNames, just not the bindings for namespaces
Aleksei: We want to avoid clashes between grammars. If we use qnames in ixml, it will fix the prefix in the grammar
<cmsmcq> If Iām reading the spec correctly, colon is not currently allowed in names.
Steven: You may be right
<cmsmcq> Adding it will require some rewriting, to prohibit colon as a final character (if we want to match the rule in xml-names).
Tom: As long as there are bindings, it's ok.
MSM: Let's agree on +
ā¦ postpone namespaces
Tom: Agree
RESOLUTION" adopt 3.2 "Use a + for inserting text"
Resolution: adopt 3.2 "Use a + for inserting text"
Resolving ambiguity
Steven: You can always fix it by writing an unambiguous grammar.
MSM: We have closed this issue. Steven has an action.
MSM: If there is more than one tree, the processor can return any one and is not obliged to explain its choice
Next meeting
6 July 2021 usual time