- From: James P. Salsman <bovik@best.com>
- Date: Fri, 31 Mar 2000 16:53:55 -0800 (PST)
- To: ietf@ietf.org, www-html@w3.org
- Cc: correspondent-address-supressed@bovik.org
Dan, Thanks for your questions: > Why are you so intent on getting this put into W3C specs, > and into implementation on user agents? My goal has never been to get published in any particular organizations' specs; only to get the major browsers to implement microphone upload suitable for language instruction, on all feasible platforms. While I was working within the W3C, I didn't realize how much Microsoft and Intel were benefiting from the non-interoperable OBJECT/EMBED tags (normative or informative, platform dependence is antithetical to the stated mission of the W3C.) And even if I had brought it up then, I have no reason to believe it would have done any good; I was in enough trouble as it was with those who seem to believe that the ability to specify pre-selected defaults is good user interface design. The only public W3C argument for that position amounts to -- and this is a direct quote -- "any proposal that suggests markup that includes the word 'device', ... should ring alarm bells." My opinion on that matter is that any user interface rubric that generalizes markup form and style over function and substance should ring louder alarms. An affirmative stance taken by the W3C would be preferable for many reasons, but the IETF's 'text/html' registration could do in a pinch. I'm not willing to make that formal proposal without asking the W3C some more questions and giving them more time to consider the aspects of the device upload specification that they have never addressed in public at all, and in most cases have not addressed even within their members-only HTML Working Group discussion areas: First, the Content-device header, which would enable a server to tell whether a picture, for example, came from a scanner or a camera, or whether a sound file came from disk or a microphone. Second, the Content-type-alternates header, which would allow for some rudimentary content negotiation, for example sending a jpeg when the gif format was unsupported or vice versa. Third, the MAXTIME attribute, which would allow compressed media with nonzero durations (e.g., sound and video but not images or text) to be limited by a maximum number of seconds instead of the less useful maximum number of bytes. Forth, the various security considerations in the draft: http://www.bovik.org/device-upload.html#Security Fifth, there is the matter mentioned above, regarding the selection of a default input DEVICE, upon which the HTML Working Group chair and I have agreed to disagree. This is a vital issue because of the legacy implementations that browsers have chosen for the ACCEPT attribute, using it as a filename pattern instead of a list of MIME types. The presence of the DEVICE attribute would allow them to make a clean break from that legacy interpretation of ACCEPT, removing one of the claimed barriers to implementation. There are other minor issues too, but the first two above are very important because they were brought about by the specific requests of the IETF Application Area during the IANA Content-type header registration process. The W3C has repeatedly refused to accept revisions to the original flawed NOTE -- http://www.w3.org/TR/device-upload -- in accord with those recommendations from the IETF, even after revisions were submitted by another of the W3C's own members. If the W3C continues to refuse revisions to multipart/form-data header parameters explicitly rejected by the IETF, why shouldn't the IETF make a corresponding amendment to the text/html MIME type registration? > What's wrong with using other applications that can do > exactly the same thing? That's the heart of the matter: What will people actually use for educational applications? The best implementations of the best specification aren't worth the electrons they're written on unless they actually benefit educational applications such as spoken language instruction, along with asynchronous audio conferencing, transcription from dictation, and the high-quality transmission of device input under low-bandwidth conditions. That last problem would benefit more of the decision-making Internet community than any educational application -- imagine! high-quality async voice messaging on your portable wireless box!!! -- but is not the source of my passion for the last remaining anytime/anyplace learning hurdle: spoken language instruction. Cheers, James
Received on Friday, 31 March 2000 19:54:13 UTC