See also: IRC log
<trackbot> Date: 13 April 2011
-> http://lists.w3.org/Archives/Member/member-poiwg/2011Apr/0024.html Agenda
<scribe> scribe: Matt
-> http://www.w3.org/2002/09/wbs/45386/POI-F2F-2011-2-choices/results F2F poll results
Andy: Not much support for Asia
meeting. Most popular was collocated with OGC and OSM in
Boulder. Second was to meet in Hungary with OMA.
... Only problem with Boulder was that it's not until end of
September.
... If we do Boulder we need a meeting in between. Maybe have a
second poll offering Denver for our 3rd and setup a f2f in
between in Europe or US as there wasn't much support for
Asia.
Fons: What about going to
Budapest meeting?
... It's much before September and is a very good option.
ahill2: Isn't Budapest one of the most popular choices?
Andy: Yes.
ahill2: It looks like Budapest was second choice.
Andy: It is number 2 amongst rank 1.
ahill2: I can't go.
Andy: We could do: Boulder and Budapest, or Boulder and TPAC. I don't think it makes sense to do TPAC and Boulder based on them being right on top of one another. Could have something informal at TPAC.
matt: TPAC would be good to present, and it would be good to meet with other WGs.
<scribe> ACTION: Andy to find meeting location space in Budapest [recorded in http://www.w3.org/2011/04/13-poiwg-minutes.html#action01]
<trackbot> Created ACTION-60 - Find meeting location space in Budapest [on Andrew Braun - due 2011-04-20].
Andy: So let's plan on Budapest and Boulder and present at TPAC.
<ahill2> +1
<jens> +1
Andy: Will we be recharterting around TPAC?
matt: Maybe just after.
ahill2: The favorite time seems to be Thursday at 10am EST.
-> http://lists.w3.org/Archives/Member/member-poiwg/2011Apr/0011.html Thread on call times
ahill2: Carl Reed said he could only attend every other week, but it sounds good. I asked for suggestions for other times and no one said anything.
Andy: Anyone have additional times?
Ronald: Thursday works fine for us.
fons: Not a problem for Europe.
ahill2: We'll do a poll to decide
between this time and the new time. If there's a flurry of
activity around it, we'll see if they suggest a third time.
It's not an urgent matter.
... Let's do a third option of a write in.
... Though maybe that doesn't get us a good response.
... but since we haven't had many responses, let's do
that.
<scribe> ACTION: matt to write up new call time poll with current time, Thursday at 10am and write in [recorded in http://www.w3.org/2011/04/13-poiwg-minutes.html#action02]
<trackbot> Created ACTION-61 - Write up new call time poll with current time, Thursday at 10am and write in [on Matt Womer - due 2011-04-20].
Andy: Reminder: please update your actions! Send mail to the list if you need help, etc.
ACTION-61?
<trackbot> ACTION-61 -- Matt Womer to write up new call time poll with current time, Thursday at 10am and write in -- due 2011-04-20 -- OPEN
<trackbot> http://www.w3.org/2010/POI/track/actions/61
-> http://www.w3.org/2010/POI/track/actions/open Open Actions
ISSUE-9?
<trackbot> ISSUE-9 -- Is an ID required within a POI itself? -- raised
<trackbot> http://www.w3.org/2010/POI/track/issues/9
ahill2: At the end of the f2f, we
were going through the meeting minutes looking for issues and
we had discussed that time and category are not required.
... Making anything required is kind of a non-starter.
... These issues are here for the group to check-off on and
agree upon that these are not required.
... Ronald had some regrets that they hadn't forced IDs.
Ronald: Basically we used to have
an ID field that wasn't required, but once we extended it for
more features, it became more difficult when POIs didn't have
unique IDs.
... If you want more dynamic usage, you probably at least want
to highly recommend it.
ahill2: Can you give us a use case? Update? Someone creates content and then there's no handle to it?
Ronald: We extended our API to
give more timely updates, use in more game-like layers.
... Might not want to resend all of the data again.
ahill2: We had a similar problem, but can't you create unique id's under the hood?
Ronald: For some cases yes. But the developer is really kind of outside of our system. Auto-generated IDs doesn't necessarily solve it.
ahill2: Is it possible in that scenario that someone could not populate the field but be assigned an ID and then they could refer back to that?
Ronald: It doesn't really solve the issue, as the ID is still there, just system generated.
fons: Yes, then it would just mean the system has to generate the ID.
ahill2: At some point we'll talk
about label. I can't help but wonder if we didn't have a name
or an ID that you'd have some real trouble in a practical
situation.
... Is it possible that this could be an implementation
specific restriction? Meaning that for your system you need the
ability to set IDs and enforce that users create them, then
does it make sense that a data format for POIs require an
ID?
Ronald: I think I could live with
that.
... It puts some restrictions on the type of communication
channels a bit. For instance if a document contains all POIs
for ever, then yes, you don't have to require ID. So
recommended but not mandatory.
ahill2: Isn't this analogous to the current Web practices?
jens: True. What we see currently
is that you write a page and then later want to give a
reference to it, you then have to go back and add an ID.
... But, I can also see a use case where providing IDs is
superfluous or too much trouble.
ahill2: There's been some
discussion about possible change in our focus to more of a
service architecture for POIs. Meaning that if there's no
registry or service architecture to deliver them then it may
not be of value.
... How does that change our view of IDs?
... In other words having a document vs a global way to refer
to them.
... If you don't have a URI and an ID there's just no getting a
POI in a Web architecture. Any comment on that?
Ronald: Is a POI only a POI if it is available in a reference system or is a POI a POI itself?
ahill2: Sure, POIs can exist,
people will put them in a file and click on them locally and
they probably won't have an ID, but does anyone have anything
intelligent to say about how a service architecture and
building one and specifying one would change this view?
... I lean towards we can't really require it.
<Andy> +1 to matt
<Ronald> +1
<fons> +1
matt: Seems to me that if you're looking at building a standalone document you don't need IDs. If you're building a linked format in a Web architecture then you do. We're doing both, so it maybe should be a SHOULD rather than a MUST.
jens: I suggest we go with not
requiring IDs, but having a recommendation to readers of the
spec that having IDs is a good idea in many situations for
future proofing your data and such. Advice on using an ID if
you can.
... "If you are making your data available use an ID"
fons: Might be a good argument to say that it might help performance.
jens: Yep!
PROPOSED RESOLUTION: IDs are not required, but are recommended. The spec will provide some advice on situations where they should and should not be used.
<scribe> ACTION: jens to provide text about the virtue of using IDs [recorded in http://www.w3.org/2011/04/13-poiwg-minutes.html#action03]
<trackbot> Created ACTION-62 - Provide text about the virtue of using IDs [on Jens de Smit - due 2011-04-20].
ISSUE-10?
<trackbot> ISSUE-10 -- Should we require a label for POIs? -- raised
<trackbot> http://www.w3.org/2010/POI/track/issues/10
ahill2: At the F2F we resolved to rename 'name' to 'label'.
-> http://www.w3.org/2011/03/31-poiwg-minutes.html#item03 F2F minutes on label
ahill2: With labels being textual
descriptions (rather than IDs) are these an option where you
can imagine a situation where someone is rendering an image and
doesn't care if there's a label. Basically a POI is some place
in the world and some thing, which could be a label or a model
or an SVG or whatever.
... You could argue that it is not required, but like IDs if
you take it out of your system, you look at a list of POIs, and
there could be a bunch that have no names. That obviously hurts
the usability. It's similar to not having an ID. At times it
will matter, and at times it won't.
jens: Then I propose a similar
resolution as we used for IDs.
... Basically if there is no virtue to having a label don't use
it. So, not a requirement, but recommended in many many use
cases. As advice to the reader, even if they aren't being used.
It is similar to alt attributes in HTML.
<ahill2> +1 to jens comment
<Ronald> +1
jens: I think the same argument
as alt labels. If you can provide meaningful text, you
should.
... Have text explaining the virtue of labels.
<fons> +1 robots can search for them as well
<scribe> ACTION: ahill2 to put text in the draft summarizing the discussion of labels thus far [recorded in http://www.w3.org/2011/04/13-poiwg-minutes.html#action04]
<trackbot> Created ACTION-63 - Put text in the draft summarizing the discussion of labels thus far [on Alex Hill - due 2011-04-20].
<jens> good suggestion fons
+10 :)
<ahill2> yes
Andy: So this closes those two issues. Let's pick another one to move on to.
ISSUE-2?
<trackbot> ISSUE-2 -- Organisations versus places: should POI core spec say how to distinguish these? -- raised
<trackbot> http://www.w3.org/2010/POI/track/issues/2
<Andy> http://www.w3.org/2010/POI/track/issues/2
ahill2: This is concepts vs places/things?
Andy: That was my impression, vs ISSUE-3 which is more about families of POIs.
ahill2: These are somewhat similar to me. About the scope of what we are describing.
-> http://www.w3.org/2011/03/29-poiwg-minutes.html#item04 Origin of ISSUE-2
ahill2: Does this spec describe
things that are not physical? Legal entities? Government
bodies? They're not physical, but are they POIs?
... Do they violate the fundamental thing about "some
information linked to some physical location in the world?"
Andy: I would say it does violate
that.
... But I think I do support this sort of structure in the POI
spec. Just from a relationship standpoint. Maybe we need to
answer some relationship questions first.
ahill2: There's also the argument
about unknown locations. We've talked about POIs if they can't
get narrowed down to WGS84, that they're at least on the
earth.
... It is possible that you could assign locations to things
that aren't very meaningful, but given relationships, there
would be value in it.
... A lot of this is facilitated in some way by categories. If
you have schemes then you have a form of relationships -- but
this ends up going down a slippery slope of describing
everything.
Ronald: I think "describing everything", as long as it has a way to tie to the real world, could be part of the POI spec, but ??? is out of scope. Then we would look to the semantic web to describe those kind of things.
ahill2: I could imagine you
wanting to link to concepts, e.g. McDonalds. You might want to
link a bunch of physical McDonalds locations to that. Is that
concept a POI?
... Does it have to be in order for the linkages to be made?
Probably not.
... There are plenty of ontologies and databases built to
describe concepts. They don't necessarily need to be grounded
in a physical location. They will need to persist and have
unique IDs, but do they have to be a POI? I don't think so.
<Andy> +1
Ronald: I think we're on the same
page. We should allow links to other kinds of data, not just to
POIs.
... I think concepts are not POIs.
ahill2: Let's not resolve this on this call, but that restriction of POIs as something that you would want to locate and tie to something geospatial in nature -- that may be something we want to resolve. That would give us some clarity moving forward.
Andy: Let's start an email thread about it and talk about it at the next meeting.
ahill2: That sounds like a good idea from a strategic standpoint. I think sometimes people think that the minutes require them to consume the whole meeting. i think it is valuable to break the issues out into email threads.
<Andy> +1
<Andy> Andy will take it
andy: I'll take an action item for this.
Andy: We've had a few resolutions
and actions.
... Just wanted to mention that the Geo group has a new
charter, check it out:
-> http://www.w3.org/2008/geolocation/charter/charter-2 Geolocation charter
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/jens/fons/ Found Scribe: Matt Inferring ScribeNick: matt Default Present: +1.919.439.aaaa, Matt, Andy, +31.20.592.aabb, Ronald, Fons, jens, +1.773.575.aacc, Alex Present: +1.919.439.aaaa Matt Andy +31.20.592.aabb Ronald Fons jens +1.773.575.aacc Alex Regrets: Jonathan Raj WARNING: No meeting chair found! You should specify the meeting chair like this: <dbooth> Chair: dbooth Found Date: 13 Apr 2011 Guessing minutes URL: http://www.w3.org/2011/04/13-poiwg-minutes.html People with action items: ahill2 andy jens matt WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]