Re: Minority opinion on resolution of issue 112.

At 2000-01-25 13:56-0500, Ian Jacobs wrote:
>Hello,
>
>I am sending this email to document the (minority) opinions held by
>Charles McCathieNevile, Gregory Rosmaita, and Ian Jacobs about
>the resolution of issue #112 [0].
>
>ISSUE:
>User agents must make known the current input configurations
>        (e.g., keyboard bindings) to the user. Is there a need
>        to distinguish author-supplied input configurations
>        from user preferences?
>
>WG RESOLUTION: Yes. P1 for user-supplied configs, P2 for
>author-supplied.
>   This resolution is reflected in checkpoints 10.1 and 10.3 of the
>   21 January 2000 UA Guidelines [1]. The resolution was made on
>   20 January 2000 [2].
>
>MINORITY OPINION:
>We do not agree that it is useful to distinguish between input
>configurations (keyboard configurations, voice configurations, etc.)
>that are supplied by the author and those set by the user. To the
>user, it is only the end result that matters. The answers to the
>following questions should not depend on whether the configuration
>came from the author or the user's preferences:
>
>  -  How does my user agent work?
>  -  What input commands are available to me?
>  -  What happens when I activate them?
>
>Some members of the Working Group have argued for two checkpoints
>of differing priorities for these reasons:
>
>   1. the UA doesn't know what to expect from author-defined
>      UI controls, or necessarily what to do with them or where to
>      look for them
>   2. author-defined UI controls may conflict with OS and UA
>      UI controls / keybindings
>   3. documenting/exposing/supporting author-defined UI
>      controls could be accorded a lower priority than UA-defined
>      UI controls
>
>All of the above objections have been addressed by members
>of the UA WG, and techniques to expose, trigger, and provide
>a cascade order for conflicting commands have been reviewed
>by the WG, and incorporated into the Techniques document.
>
>Since ultimately the user agent is the only entity (not the author,
>not the user) that "knows" what the input configuration is
>(after taking into account user preferences and author-supplied
>configurations), we consider that it is the user agent's responsibility
>to make that information known to the user. We see no reason
>to distinguish in priority between author-supplied configurations
>and user preferences because, to the user, they have the same weight.
>
>Thank you,
>
>  - Ian

HB: I agree that it is the UA's responsibility to let the user ask and learn
the full content of the input controls. But, if the user doesn't give a 
complete specification of all possible input controls, then surprises are 
likely
for any other controls that an author may change from the OS default.
For that reason, I believe there are two distinct checkpoints, unless
the result of user query about who established each control is multi-valued:
user profile, user agent default, AT default, OS default, or author override.

A further concern is that there is no presumption that the set of controls
is consistent across OS, user agent or AT running under it, or application
in which an author presents information (HTML, XHTML, any XML application).
Cross-OS use of a profile is unlikely unless the members of this set of
controls is fixed and have stable, consistent definitions.

Regards/Harvey Bingham

Received on Tuesday, 25 January 2000 20:15:04 UTC