> I believe file submission is mainly a browser issue -- one should be
> able to (say) drag-n-drop a file into any TEXTAREA and have the
> contents POSTed as the field's value. The author of the page cannot
> (and perhaps should not try to) anticipate in which fields the user
> might want to place the contents of a file instead of typing by hand.
While this might be reasonable if you're thinking the submissions as
mainly text, it isn't reasonable to have a TEXTAREA for uploading a
postscript file, or audio, or graphics. Nor is it particularly useful
if the input is allowed to be multiple items.
In addition, the upload method for large data streams cannot usefully
be done with "x-urlencoded" or whatever.
> Your [nebel@xsoft.sd.xerox.com (Ernesto Nebel)] proposal does provide
> some additional features that could be of use to server maintainers.
> On the down side, in addition to requiring changes of both browsers
> and servers,
It only requires changes to servers that want to accept data in this
way, and those changes seem to be accomplishable with CGI scripts;
it *does* require browser extensions, but as has been pointed out,
it can be handled in a way that current browsers will function, just
not optimially.
> it would posit some rather fundamental changes to the
> current model of HTTP as a stateless protocol.
There's nothing in the proposal that requires *any* change to the HTTP
protocol, so this is a misstatement.
> I don't think such a change should be made lightly.
I agree. I'll observe, though, that the proposal wasn't made lightly;
it was posted on this and other mailing lists, and suggestions for
improvements or alternatives solicited. It's certainly a widely
requested feature, and a prerequisite for useful deployment of a wide
variety of web-based services.