Re: Device upload for all platforms -- the official HTML WG position

Dear Dr. Pemberton,

Thanks for your message:

> There is nothing in HTML 4 that excludes any platform.
> Just look at Opera, which is being implemented on BeOs, Epoc, Linux, 
> Mac Os, OS/2 and Windows.

Device upload -- of any kind -- has not yet been implemented in Opera.  
You know that the CTO of Opera software has said they will wait for a 
W3C Recommendation (or Working Draft) on device upload.  Opera, even 
on Windows, will not interpret MSIE OBJECT or Netscape EMBED tags which 
are used for microphone upload on those browsers under Windows.  There 
is simply no browser on Mac or Linux (other than my experimental 
build of pre-gecko Mozilla) which is capable of even microphone upload.

> If you had listened to the comments to your proposal when you submitted
> it, you could have corrected it, and it would be a non-issue.

Since 1997, there have been eight revisions, incorporating the useful 
suggestions from anyone who cared to review and comment on the draft.  
The basic idea and method remained the same throughout; the significant 
changes mostly concerned the names of devices.

> In your case we pointed out there are fundamental flaws in your 
> approach, and indicated how you could fix them.

I am sure we both want to resolve this.  Would you please list all 
the flaws of which you are aware -- with as little or as much detail 
as you have time for -- along with, when available, how they could 
be fixed?  I promise you I will devote my efforts to your comments 
until you are satisfied that they are resolved, but I do need your 
help to understand which issues you consider flaws.

I will start by addressing the question of whether ACCEPT alone is
sufficient for device upload, when used with the INPUT TYPE=FILE 
element.  The problem occurs when an image is requested on a system 
with both a camera and a scanner.  Having a DEVICE attribute allows 
for the selection of a default, which is reasonable because some 
conferencing applications might always expect camera input whereas 
OCR applications would usually expect scanner input.  And, as others 
have pointed out, in compressed media the MAXLENGTH limitation is 
not as practical as the MAXTIME limitation, in the case where an 
upload needs to be restricted in size.  Does anyone think that this 
justification is inadequate?


Received on Wednesday, 1 March 2000 14:33:30 UTC