Forms Working Group Teleconference

04 May 2011

See also: IRC log


+44.782.483.aaaa, Leigh_Klotz, pfennell, alain, Nick_van_den_Bleeken, John_Boyer, unl, ebruchez


<trackbot> Date: 04 May 2011

<ebruchez> All, sorry I will call a bit later today

ok ebruchez irc only for now.

<unl> me too, phone set broken again

<nick> scribe: nick

"Model-driven Views" thread on webapps

<klotz> http://lists.w3.org/Archives/Public/public-webapps/2011AprJun/0406.html

klotz: Someone from google re-invented XForms

<klotz> Charles McCathieNevile says "That in turn was a pretty straight rip-off from Xforms which allows the

<klotz> same thing with a little more power, and is even closer to what it seems

<klotz> you're looking to do."

klotz: We should ask to talk with us
... anybody against us contacting them
... maybe we should keep our response general
... I will send the response if that's ok

<scribe> ACTION: klotz to repond to http://lists.w3.org/Archives/Public/public-webapps/2011AprJun/0406.html [recorded in http://www.w3.org/2011/05/04-forms-minutes.html#action01]

<trackbot> Created ACTION-1796 - Repond to http://lists.w3.org/Archives/Public/public-webapps/2011AprJun/0406.html [on Leigh Klotz, Jr. - due 2011-05-11].

<klotz> http://lists.w3.org/Archives/Public/public-forms/2011Apr/0019.html

transform function module

klotz: philip for the people on the phone your the most involved can you sumerize?

pfennell: Under my point of view it shouldn't be a function but an action, so the transform could be in the instance

nick: what is the difference between the function and the action concerning this

John_Boyer: with the function it is difficult to know if it is a uri or a node
... When it is in the instance you have depency problems because currently we don't add dependencies to the descendants

nick: we have that problem also with id() function and some others, and said then that we could solve this by adding extra dependencies

John_Boyer: we didn't solve this on the wiki because we sidestepped the issue by only allowing uri's

pfennell: can't you use #id?

nick: you can't point inside the live instance
... you can only reference the static document which is present at load time
... we can't point to the live instance

<klotz> instance:#xpointer(instance('foo')/bar)

instance: //id-of-instance#id could work too

klotz: we could define duration of the session to fetch the url's
... in XSLT they define that the url's are de-referenced only once during a transform

nick: It is the same problem as we discussed for XPath 2.0

John_Boyer: We could determine the type based on the type of the node (if it is a URI)

klotz: We have to be careful that it also works for other transform language


like XQuery

klotz: Three case:

1) URI

2) it is top level element allowed by xslt

3) If it is string

<alain> Sorry, I have to leave now... Bye!

John_Boyer: I thought sting would default to URI

klotz: then it won't work for XQuery
... we could add an extra parameter pointing out the parameter value

nick: I don't like the idea of parameter that dynamicly identifies the parameter type

<John_Boyer> transform(object, objectType, inputNode?). objectType would say how to interpret the object, xslt or xquery or what have you.

<John_Boyer> maybe a mistake, but the actual type analysis on the first parameter could be used to decide whether it provides a URI or whether it provides the octets of the transformer

klotz: If we have support for multiple transform languages we can rely on the media type for a uri
... but this doen't work for nodes

John_Boyer: what about file:// protocol
... then you don't have media-types

klotz: they could determine the transformer type

<John_Boyer> objectType => mediaTypeOfObject

klotz: for the local objects it will be much harder to implement

nick: #id is also a pointer to a node

<John_Boyer> identifying the type of the content referenced by the first param object is a separate problem from identifying the type of the first param object

klotz: we could go for two functions

<John_Boyer> Erik is asking about the type of the first param object, being just URI or also being a node whose string content *is* the transformer

klotz: we could say that it is nver a URI

nick: In my use cases I don't need to change the stylesheet dynamicly and therefor I would keep my stylesheets externally

John_Boyer: We could define the implementation of the function, so we could even in XPath 1.0 ask the processor what the type is, and if it is derived from anyURI we could dereference to URI

<John_Boyer> <bind nodeset="some/node" type="xsd:anyURI"/>

(discussion about what a valid URI is according XSD)

John_Boyer: what is the mime-type of XSLT

nick: isn't it just application/XML

<John_Boyer> transform(some/node) would treat the node as a URI because that is its type, and this could be done in XPath 1.0

<John_Boyer> challenges: default is string, which seems wrong

John_Boyer: We have a challenge on the first parameter because the default should be string and it would be anyURI for this

<John_Boyer> and for the second arg, the challenge is that there aren't good mediatypes for the transforms, e.g. xslt is just application/xml

klotz: We don't want to use type polymorphism for the first parameter so we suggest two parameters

ebruchez: If you do two functions how do you handle the XQuery case

klotz: if it is a uri the de mime-type would be application/xquery

ebruchez: historically you specify a type in html, of course there is default

klotz: why do you come back with a proposal?

<John_Boyer> also a problem with current tranform() because the first param URI will return octets whose mediatype won't clearly identify xslt

John_Boyer: There is something wrong with the currently wrong with the transform function in the wiki because it doesn't has a media-type parameter

pfennell: another transformation language could be XProc

<klotz> The media type application/xslt+xml will be registered for XSLT stylesheet modules

<John_Boyer> wouldn't want to have to parse the application/xml content in order to get the root element namespace in order to decide if it is xstl vs. xproc vs. some other transformation.

<klotz> scrobe:klotz

<klotz> scribenick:klotz

<scribe> ACTION: ebruchez to propose new name for transform for uri in email. [recorded in http://www.w3.org/2011/05/04-forms-minutes.html#action02]

<trackbot> Created ACTION-1797 - Propose new name for transform for uri in email. [on Erik Bruchez - due 2011-05-11].

Summary of Action Items

[NEW] ACTION: ebruchez to propose new name for transform for uri in email. [recorded in http://www.w3.org/2011/05/04-forms-minutes.html#action02]
[NEW] ACTION: klotz to repond to http://lists.w3.org/Archives/Public/public-webapps/2011AprJun/0406.html [recorded in http://www.w3.org/2011/05/04-forms-minutes.html#action01]
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/05/04 16:04:00 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
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/media-type/mime-type/
Succeeded: s/default is string/default should be string/
Succeeded: s/should be anyURI/would be anyURI/
Found Scribe: nick
Inferring ScribeNick: nick
Found ScribeNick: klotz
ScribeNicks: klotz, nick
Default Present: +44.782.483.aaaa, Leigh_Klotz, pfennell, alain, Nick_van_den_Bleeken, John_Boyer, unl, ebruchez
Present: +44.782.483.aaaa Leigh_Klotz pfennell alain Nick_van_den_Bleeken John_Boyer unl ebruchez

WARNING: No meeting chair found!
You should specify the meeting chair like this:
<dbooth> Chair: dbooth

Found Date: 04 May 2011
Guessing minutes URL: http://www.w3.org/2011/05/04-forms-minutes.html
People with action items: ebruchez klotz

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]