IRC log of wmca on 2004-10-13

Timestamps are in UTC.

10:36:46 [RRSAgent]
RRSAgent has joined #wmca
10:38:24 [maxf]
maxf has joined #wmca
10:38:37 [maxf]
10:45:47 [wendy]
xya, nokia: related to xforms?
10:45:53 [wendy]
ls yes. it's a fantastic accessible platform.
10:46:12 [maxf]
xya is Oskari
10:46:22 [wendy]
ls few people are building xforms. convert html forms to xforms.
10:46:57 [wendy]
ls add metadata and render as xforms. but if user agent doesn't suppport, can render as html form and reposition the text so that the label sits with the input.
10:47:14 [wendy]
ls onc eyou know that this piece of text is help for the form control, can adjust the presentation.
10:47:24 [wendy]
roland, ibm
10:47:56 [wendy]
roland: you can't use the same words/sentence structures...people admitting to be blind....admitting reading level is a bigger problem.
10:48:16 [wendy]
roland dealing with quantity of info, is a staggering problem. UK govnt, 80% of users, 20% are disadvantaged.
10:48:28 [wendy]
roland it's more than expanding acronyms, they couldn't understand the concepts
10:48:34 [wendy]
10:48:52 [wendy]
ls can't make site available to everyone. some people will not undersatnd.
10:49:10 [wendy]
ls why we need to move out of the page. can't make "a page" that is a good experience for everyone.
10:49:38 [wendy]
ls need to reach the people who can understand the concepts
10:50:01 [wendy]
ls text is a barrier to understanding the concepts. perhaps with page map, they can understand.
10:51:05 [wendy]
10:51:56 [wendy]
Uni. Dresden, Gerald Hubsch
11:08:37 [wendy]
oops. m issed the question
11:08:47 [wendy]
gh the fragmentation takes place before teh xslt is applied.
11:08:56 [wendy]
gh split up into most # of atoms.
11:09:16 [wendy]
gh wml cmoplier generate, documentvalidation validates...don't know if validates for all documents
11:09:25 [wendy]
fabio generate the markup before deciding how long
11:09:30 [wendy]
gh at run time is one option
11:09:39 [wendy]
rhys happy to see xpath.
11:09:50 [wendy]
rhys hav eyou looked at our xpath spec in content selection
11:09:58 [wendy]
rl it's worth a look
11:10:14 [wendy]
rl device classification: interest in how you do that and axes
11:10:26 [wendy]
gh for the protype it's straightforward. mobile, pda, notebook
11:10:38 [wendy]
gh othe rapproaches used in rendering language devloped in consensus project
11:10:59 [wendy]
gh use device classification that divides into 7 classes depend on input/output capabilities, voice capable, etc
11:11:12 [wendy]
rl we know aobut that project, just wondering
11:11:37 [wendy]
rl my company does this commercially, the fragmenation is about device capability and not about author choice.
11:11:53 [wendy]
rl you might find mileage about putting it outside of content
11:12:03 [wendy]
rl diff splitting/device could get you some mileage
11:12:19 [wendy]
gh agree that keep structure info about semantically related from the content is more effective
11:12:42 [wendy]
rh if using a tree there is inherent semantic interpretation in traversing tree is same path as dialog
11:12:49 [wendy]
rh in some cases, more appropriate to change the order
11:12:56 [wendy]
rh in your scenario, can't do that
11:13:08 [wendy]
gh split it up into components? ya, ok
11:13:16 [wendy]
rh what use for fragment url?
11:13:48 [wendy]
gh after fragmentation, those fragments cached for sequential delivery to the client, after the client has finished it decides if whoel doc or part of doc has been fetched.
11:13:57 [wendy]
gh decides which fragment to deliver next
11:14:02 [wendy]
gh delivers from cache
11:14:10 [wendy]
rh what is the tech used for fragment uri. just id?
11:14:12 [wendy]
gh just id
11:14:30 [wendy]
rh it's a resurfacing debate about how to reference something in a document
11:14:46 [wendy]
11:15:08 [wendy]
gh there are options to include layout information, specific properties. filte rproperties out for diff devices
11:15:15 [wendy]
11:15:43 [wendy]
BT, Tim Glover
11:33:39 [wendy]
icra, at what point did you think metadata would be useful?
11:34:03 [wendy]
tg html itself is not data, it has suggestions for how to lay things out. to me, that's metadata.
11:34:21 [wendy]
rh this shows how far you can get w/out metadta. gives a hint for essential metadata.
11:34:34 [wendy]
ah what is approach thinking for accessibilit preferences? recommend IMS AccLip
11:35:02 [wendy]
bb how do you select a template?
11:35:19 [wendy]
tg conditional elements w/in template. create a template for each uri, in order. first will be selected.
11:35:28 [wendy]
bb those conditions...isn't that metadat?
11:35:38 [wendy]
tg based on device profile
11:35:57 [wendy]
tg simple boolean. want more sophisticated
11:36:02 [wendy]
bb like media queries?
11:36:05 [wendy]
tg yes, like
11:36:22 [wendy]
sb what did you say about expression querying profile? in doc, in server?
11:36:39 [wendy]
tg each template associated with a condition. outside condition, can associate templates.
11:36:54 [wendy]
sb have a file w/all templates?
11:37:01 [wendy]
tg yes, all tempaltes, mimetypes, and conditions
11:37:10 [wendy]
sb exploring in diwg.
11:37:15 [wendy]
tg it has include built in
11:37:21 [wendy]
sb based on spec or your own?
11:37:32 [wendy]
tg i can say %include [file]
11:37:36 [wendy]
tg server-side include
11:37:56 [wendy]
xyz i've written lots of xslt, however this approach seems to suffer from same thing as xslt. very specific for a specific page.
11:38:02 [maxf]
11:38:05 [wendy]
.. how reusable are they?
11:38:17 [wendy]
tg suggest we use sql, it is much more abstract
11:38:35 [wendy]
tg for small apps, you have a stable dtd, you can write templates that match.
11:38:57 [wendy]
tg personally, feel that a relational database is superior way. sql is supier to xquery.
11:39:17 [wendy]
, orange: i wrote a book on xml and sql. ought to read.
11:39:43 [wendy]
roland concept-dervice characteristics should be?
11:39:48 [wendy]
tg thought about vaguely.
11:40:06 [wendy]
tg capabilities classes more appropriate than device characteristidcs.
11:40:11 [wendy]
tg could write OWL to write
11:40:25 [wendy]
tg small screen is one example
11:40:40 [wendy]
11:44:20 [wendy]
RRSAgent, make team
11:44:20 [RRSAgent]
I'm logging. I don't understand 'make team', wendy. Try /msg RRSAgent help
11:44:27 [wendy]
RRSAgent, make team visible
11:44:27 [RRSAgent]
I'm logging. I don't understand 'make team visible', wendy. Try /msg RRSAgent help
11:44:37 [wendy]
RRSAgent, make log team
11:44:42 [wendy]
zoinks, i'm tired...
14:56:23 [RRSAgent]
RRSAgent has joined #wmca
14:56:48 [wendy]
14:57:01 [wendy]
wrote down additional thoughts
14:57:09 [wendy]
looked at metadata that might not be in the content space
14:57:36 [wendy]
way to classify thousands of devices
14:57:44 [wendy]
a rough notion of how look on diff devices
14:58:02 [wendy]
what would it mean to have content-side metadata that said "what was the author thinking about this class of devices when built this..."
14:58:21 [wendy]
debated that probably not just devices, may need to cover something to do with user modeling
14:58:26 [wendy]
what is the user likely to do with this stuff
14:58:33 [wendy]
a topic for diwg to discuss
14:58:59 [wendy]
what you like if you like large text, here's for bigger text
14:59:04 [wendy]
work across devices and user profiles
14:59:10 [wendy]
14:59:20 [wendy]
role seemed to be key thing
14:59:45 [wendy]
clearly a need to say, "it has a role that i can't express alone in xhtml"
15:00:11 [wendy]
also talked about tasks. concluded that task, an achievement th euser makes on the site, could also be a role
15:00:19 [wendy]
roles could span multiple pages, not just defined in one page
15:01:01 [wendy]
adaptation rules - what need to say?
15:01:20 [wendy]
image degredation - scaling down? clipping? diff kind of data (text?)
15:01:27 [wendy]
this at principles level: general ways to do transformation
15:01:43 [wendy]
also general principels for metadata itself
15:02:55 [wendy]
scalable, maintain it further downstream than you think (perhaps not metadata itself, but references to it), standards-based, if extend - do in a findable way
15:03:05 [wendy]
took some of the things we came up with yesterday, identified some as key