Meeting minutes
/www.w3.org/2025/03/18-ixml-minutes.html///www.w3.org/2025/03/18-ixml-minutes.html
Accept the minutes of the previous meeting
[Agreed]
Review of open actions
[None]
Status reports
Steven: I'm writing a proposal for ixml modularisation for MarkupUK
A subtraction operator
Steven: I would prefer a topic title that includes the exception operator
Steven: So John, you think subtraction would cover more use-cases?
John: We need an efficient method of saying "Unless it's a keyword"
… many examples talk about subtraction in the tokenisation phase.
… it was comparitively simple to implement in Earley for me
… MIchael used to provide examples that it wasn't good for.
… there are other mechanisms to do it
… but they are ugly, using character by character.
… but it won't work for certain edge cases
… and possible not with other parsing algorithms.
Steven: The ! construct would allow things like regexp-style behaviour that people are calling for.
Fredrik: Both might be relevant
… I've been implementing the ! construct, difficult since I didn't implement the parser myself. ! is interesting because you can use ! at the end of a rule for instance.
… it seems like the two solve slightly different problems
[Fredrik demonstrates]
Steven: Why do you restrict it to literals?
Fredrik: Lookahead
Steven: I see a way of doing it with general nonterminals
Fredrik: What if the guard is longer than the thing it is guarding?
John: subtraction would work with nonterminals
… I have demonstrated that it can work and match not a substring, and it certainly works where the RHS is any finite sized structure
… it has problems with zero-lengths
… I compile the LHS and RHS of the accept operator into separate rules,
… and we are looking for the completion of those nontermianls
Fredrik: Can you put it at the end of a rule?
John: Yes
Fredrik: I don't think the binary except operator can be used at the end of a rule.
John: True.
… you might be able to construct the one from the other.
… if either LHS or RHS of the except operator is the null string, it's difficult
… the LHS it works OK; see the issue
Steven: Interesting. I see the next step as comparing the use cases.
Nico: I see tokenisation as a separate issue.
… I have hundreds of thousands of tokens.
… I plan to write a paper about it.
… ixml takes plain text and produces XML; that is asymmetric
… when you're dealing with XML as input, it becomes a whole new issue.
John: So you're looking at structure within structure
… XSLT may offer you help
[Explains the method]
John: You can have ixml grammars as leaves of XML documents.
… I'll write something about this in the next weeks.
Nico: I would like a summary.
Fredrik: I am experimenting with the ! operator
AOB
[None]
[Adjourn]
<Steven> s/Topic Accept the minutes of the previouhttps://