ERT WG

re-adoption of uri:uri in WebContent class

<shadi> http://lists.w3.org/Archives/Public/public-wai-ert/2006Oct/0073.html

saz: there's no need for uri:uri in public web content, but there are situations where the URI is not unique
... so earl:WebContent should have an uri:uri additionally to a unique rdf:ID (or rdf:about)
... information on how to use uri:uri and rdf:about should be in the guide

ci: WebContent is only for World Wide Web, local file system or private web servers are not world-wide
... both use cases should be handled by FileContent

saz: WebContent is for Web (HTTP), not necessarily World-Wide

cv: or FTP or ...

saz: WebContent right now is focused on HTTP (request/response stuff)

cv: can't I use WebContent for ftp resources?

saz: you can
... it's not just HTTP

saz/cv: you can use WebContent for basic FTP operations (using ftp URI)

cv: subclass WebContent and put uri:uri into subclass

ci: private stuff needs uri:uri, public stuff doesn't

cv: rdf:about doesn't work with content negotiation

ci: so uri:uri must be used then

saz: make uri:uri optional (when needed)?
... used only when needed?

ci: we need to explain the use cases when to use uri:uri and when to use rdf:about, could be tricky

cv: don't see much difficulty in explaining

saz: we're talking about software developers, they should know their way

cv: when using http:GetRequest, uri:uri is not necessary on WebContent

ci: still think it could be difficult, but go on

saz: I now think to make uri:uri mandatory on WebContent

jk: if used on both classes, it must be the same, that's the only requirement I see

<scribe> ACTION: saz to write a follow-up on when and how many times to use uri:uri [recorded in http://www.w3.org/2006/11/08-er-minutes.html#action01]

proposal for FileContent class

<shadi> http://lists.w3.org/Archives/Public/public-wai-ert/2006Oct/0072.html

saz: filename is subproperty of uri:uri

<shadi> file://index.html

jk: I wouldn't make it a subproperty, if it's not really a URI

AFAIK this means: the server host is index.html

file: ///index.html means: a file with the path index.html in the file root on the local host

ci: you cannot just put file: in front of it and think it's a proper file URI

saz: do we have a data type for file names?
... we can leave it completely open (forward/backward slash, case sensitivity)

What is the benefit of using 'file://my/path/to/index.html' instead of 'index.html'?

or even 'my discosed index.html'?

<shadi> ./index.html

jk: this has no indication of URI scheme, could be file, ftp, http, or even relative to the EARL resource containing it

<shadi> ./www/index.html

<shadi> index.html

jk: then call it file path, because it's a relative file path

<CarlosV> file:./www/index.html

cv/saz: the use case is about finding files, not just labelling them

saz: are there issues with cv's example?

The path is '/www.index.html'

at least according to the Java implementation

oops, sorry

jk: confirmed

<CarlosV> file://localhost/./www/index.html

<CarlosV> ;-)

<scribe> ACTION: cv to post his thoughts on file name [recorded in http://www.w3.org/2006/11/08-er-minutes.html#action02]