W3C home > Mailing lists > Public > public-wai-ert@w3.org > October 2006

RE: test subject is a local file

From: Carlos Iglesias <carlos.iglesias@fundacionctic.org>
Date: Wed, 4 Oct 2006 03:00:48 +0200
Message-ID: <09700B613C4DD84FA9F2FEA5218828190E7DCE@ayalga.fundacionctic.org>
To: "Shadi Abou-Zahra" <shadi@w3.org>, <public-wai-ert@w3.org>

Hi group,

>During last week's teleconference we started to discuss the issue of describing test subjects that are local files. There are two >specific issues to consider in this:
>1. How to identify the test subject uniquely (a filename may not be unique outside the local context)
>2. Is it necessary to have a mechanism to record the contents of the files? If so, how would the mechanism look like?
>The following proposals have been raised:
>A. Use the existing WebContent class even though this is not information on the Web. This would also imply changing the >description of the class in the spec as well as changing the domain of the http:body property so that it can be used for local >files.
>B. Develop a new FileContent class that contains optional properties to record the file name, its contents, as well as other >properties that we may consider. So really a custom class for local files as we did for Web content.
>C. Use the filename as the URI of a TestSubject class (neglecting that this may not be unique outside the local context), and >forget about recording the actual contents of the files.
>What do people think of these proposals? Are there other proposals?

My favourite is the B option, using the file://whatever as the URI and incorporating some way to include the content (or at least keeping it in mind for an easy later extensibility). I think there are enough differences to justify a new specialization of the TestSubject.

Can live with the A option and don't like C option at all.

Received on Wednesday, 4 October 2006 01:01:10 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 20:55:54 UTC