See also: IRC log
<scribe> scribe: Phil
<scribe> scribeNick: phila
<renato> Thanks ;-)
benws2: Any objections?
RESOLUTION: Last week's minutes accepted
<renato> https://lists.w3.org/Archives/Public/public-poe-wg/2017Jan/0012.html
benws2: renato Wouldyou like to introduce your solution to complex constraints and how that extendes to extended relations
renato: Idea is that constraints
on constraints work by allowing L and R operand to be existing
constraint objects
... This is supported in our 3 syntaxws
... use XOR or AND
... Create the 2 constraints and then say XOR between C1 and
C2
... That is then closely related to extended relations
... If we only constrained constraints... we already have
L&R operands. Doesn't fit with duties
... TNhe proposal is not to support extended relations for
duties as it doesan't have the same functions as
constraints
... where we have L & R operands. We don't have tht for
duties
... So if we only limit ERs to Constraints then we're OK
... The use case for this was a duty that had 2
constraints
... We'll support this.
<simonstey> +q
benws2: The 1st question - would anyone suffer if ER were no longer applicable to duties?
simonstey: Regarding the use
case, there's the famous example of breaking the law, you can
choose whether you want to pay or go to gaol, so you have 2
ways to act
... But another question...
... in the e-mail, renato, you said ... AIUI, each L & R
operand would be a constraint object
... So we might have dateTime for Left Op, and then ...
... So end of 2016, OR end of 2017
... Or can we only bundle constraints... (scribe lost
detail)
renato: At the constraint level, the 2 objects would be constraints
simonstey: I don't see why,
without doing the changes, I think legacy ODRL, I think those
extended relations are also easy?
... I don't see how the changes affect on a constraint
level
renato: I think before we had
explicit left operand, where the LO was an actual constraint,
say, 'day time' then it seemed harder to model the ERs between
2 separate constraints.
... Now it's clear that we have LO and RO, and they can be
constraint objects, it seems clearer
simonstey: I'd have C1 less than C2...
[Agreement]
benws2: TO summarise...
... The model doesn't change, just the syntax?
renato: Pretty much. There are no
major changes. maybe a bit of narrative updatwe for how to use
them. But there's no structural change to the model
... As I replied to Stuart, we may have processing rules for
XML
benws2: No impact on expressivity?
simonstey: You can say more
things
... AND brings no additional expressivity, but OR and XOR
do
benws2: I meant compared with the old model
simonstey: Hopefully only the
extended expressivity.
... It's really just the same thing.
smyles: I agree that the proposed
solution works, and it does make the model more powerful
... I just think we'll have to work through the details of the
syntax
... You need a slightly more powerful processing model than we
did for ODRL 2.1
... Otherwise I think it works.
benws2: On the concept of Cs on Cs...
<simonstey> +q
simonstey: The ERs are fine and
understandable
... what this depends on... it's tricky. There are many
different ways to look at it.
... Does depend have a logical implication?
... If A depends on B then can't be true if B is false, but
that doesn't imply any temporal order
... And then there is this temporal dependency: first X has to
be true, then y has to be true - and that's not the same.
... Depending how we define our ERs could extend to the
situation where you have a nested XOR on one side
... and this depends on another giant construct, both referring
to the same Cs in a loop
... When do you want to check, how long do you have to
wait?
... Many differnet things to consider and be aware of.
benws2: Are you optimistic?
simonstey: Optimistic with ERs,
but not the dependencies
... We have to be super clear on how Cs that depend on each
other have to be validated and processed.
... Maybe we have to make it impossible to point to
certain...
benws2: I think the way to progress, is to generate some cases that you can bring back to the group
action simonstey to generate some use cases the show problems with dependencies
<trackbot> Created ACTION-39 - Generate some use cases the show problems with dependencies [on Simon Steyskal - due 2017-01-30].
benws2: Does that capture what you wanted to do?
simonstey: There are many different aspects. Is this feasible, anda then describe it in a bullet proof way
michaelS: I sent a comment
earlier today... it goers close to what Simon has
outlined.
... Maybe we can trim it down to a sequence
... Then I notice that renato says something like that was
defined for ODRL1 but then taken out.
... My basic concern is that it gets too complex.
... Could there be a chain of such dependencies... C1 on C2 on
C3 etc.
... I'd like to trim it down to things that are widely
requested
benws2: So could you collaborate with Simon on some cases?
michaelS: Yes
renato: I agree with both the
previous points. It can get complicated, that's why we took it
out.
... What we could do... we could state that C1 and C2 should be
atomic constraints, i.e. not part of another expression, so you
always havea a simple operation
... So I think we should include that in the processing
model
... Happy for others to come up with better wording of
course.
... And the other thing... the other option is not to support
Cs on Cs and say it's too hard and will cause problems in the
future.
benws2: Just on Cs on Cs, we've
talked about it a lot, which is worrying, but there are some
solid use cases so it's clearly needed.
... We do have to respond to constraints that are more complex
than the model currently supports. Embargo being a simple case
in point
renato: Sure this proposal forCs on Cs we should stipulate that C1 nad C2 should be atomic
<simonstey> precedes/succeeds ?
<Zakim> phila, you wanted to talk about test suites (again)
phila: Exhorts Simon and Michael to think in terms of test data when coming up with these examples
simonstey: Seems reasonable.
Maybe we can work out a way that those sequential Cs could
work. The current version wouldn't work in long chains
... You need to know if it;s the same sequence, or a separate
sequence, I don't see how this works without URIs
... There may be a C that precedes another C etc.
benws2: Any more to say on constraints?
[Silence]
<benws2> https://www.w3.org/2016/poe/wiki/Deliverables
renato: An update..
... The vocab and model have been updated in the EDs
... Less than a day ago.
<renato> https://github.com/w3c/poe/issues
renato: All current issues,
including those ones we are talking about today...
... Most are now marked to be closed
... With a Red tag
... That will cover the approved requirements, the multiple
actions and parties...
... So I think the current state of the 23 TR docs to be -
after we make the changes after today, I think the docs will be
ready to be published as next iterations
... Ideally we can vote next Monday
... Depending whether Simon and Michael's use cases raise any
other issues along the way,
benws2: OK
renato: What that means for the
rest of the WG is: please read the current 2 specs. You can
follow through the issues, or just read the drafts
... Both docs include change logs
... Please read through and raise any issues, especially
substantive ones.
... We need to get the wide reviews so ideally we're close to
having all the functional detail done
benws2: Are we going to update the UCR?
renato: Yes, but it doesn't have to be simultaneous
<ivan> +1 to phil
phila: No rules on this, but if it is only a week before UCR is ready, then it makes a better PR story if they all go out together
benws2: OK, can my fellow UCR editors get the doc done in 2 weeks?
<renato> "a hard fortnight" ;-)
[General discussion about getting the UCR doc updated]
benws2: Anything else under deliverables?
<renato> https://www.w3.org/2016/poe/wiki/Deliverables
renato: The list of who we're
going to seek wide review from
... More names/organisations would be good there
<scribe> ACTION: phila to find example of wide review e-mail text [recorded in http://www.w3.org/2017/01/23-poe-minutes.html#action01]
<trackbot> Created ACTION-40 - Find example of wide review e-mail text [on Phil Archer - due 2017-01-30].
ivan: There are some WGs that
have their own self-testing questionnaire
... Privacy I think has that
phila: Yes, i18n does too
ivan: So it might be worth going
through that immediately after publication.
... Privacy folks will ask you to go through that before they
look at it
<ivan> http://gregnorc.github.io/ping-privacy-questions/
<ivan> https://w3ctag.github.io/security-questionnaire/
<ivan> https://www.w3.org/International/techniques/developing-specs?collapse
<simonstey> +q
phila: I don't think this WG is relevant to a11y?
ivan: Probably not, but in the past they have commented on the spec itself.
[Phil and Ivan discuss a11y needs]
<scribe> ACTION: phila Look into accessible SVG version of the model diagram [recorded in http://www.w3.org/2017/01/23-poe-minutes.html#action02]
<trackbot> Created ACTION-41 - Look into accessible svg version of the model diagram [on Phil Archer - due 2017-01-30].
<simonstey> https://www.w3.org/WAI/ER/tools/
simonstey: I can remember when we
had a more tedious method of checking docs through a11y
... If you pass those guidelines then you could go.
-> https://www.w3.org/pubrules/ PubRules
ivan: I don't think PubRules does a11y atthe moment
simonstey: I can imagine that for our vocab to be accessible, something to complain about is certain concepts may not be full described
ivan: I think having a description for each terms is a good thing
<benws2> https://www.w3.org/2016/poe/track/actions/open
benws2: Reads through the open
issues, noting his own...
... Unit of count. I remember having a discussion about this. I
don't think Good Relations provided that support.
<Serena> *you mean Sabrina I guess :)*
simonstey: I know Sabrina is busy this week
close action-34
<trackbot> Closed action-34.
benws2: I think we can close action-36
close action-36
<trackbot> Closed action-36.
benws2: We have decided on London...
close action-37
<trackbot> Closed action-37.
benws2: Any more to say on actions?
renato: At this stage it's about logistics. We'll work on the agenda nearer the time but dates and venues are decided.
benws2: I'll provide recommended hotels etc.
https://lists.w3.org/Archives/Public/public-poe-comments/2017Jan/0002.html
<simonstey> +q
benws2: Might be worth pinging
Victor directly
... And I did this translation some time ago
simonstey: I'm in Dalicc - we're
responsible for the processing, not the modelling
... They tried to read the entire licence and see whether there
were pieces missing
... to represent the whole licence is complicated
<renato> My response: https://lists.w3.org/Archives/Public/public-poe-comments/2017Jan/0003.html
simonstey: I had a meeting with a legal expert . They're trying to compare ccBy with other open licences and then we can some up with a way to crrectly represent them.
<renato> Duties are not "correctly" modelled...
phila: Mentions the work at the EU data Portal where many licences have been compared
simonstey: Not sure that listing all the actions is the right approach.
benws2: Wraps up