1. Being able to upload a file using HTTP PUT method. This is the more efficient way from the server side (no need to parse multipart separators), and probably from the client side (no need to store the file in RAM). I guess this could simply be done by allowing to have the method=”put” attribute to the HTML form. Then each form input would be sent in a dedicated PUT request, with the field name appended to the action URL, and the field value as the sole request body. (similarly, i would like to be able to trigger DELETE request from HTML, but this is out of scope of this discussion)

2. Allow a file input to upload multiple files: instead of holding a single file path, the input value would hold a list of file paths. This could be done by adding a new “fileset” input type.

I understand that these suggestions impact the HTML, and hence raise a compliance/interoperability issue. But I would appreciate Mozilla to prototype such solutions, and promote them to w3c.

It would be nice to extend the file input element (widget) with several options:

* Multiple file select. When only one file is selected, the filename.ext is shown (not the entire path, which is useless information anyway). When multiple files are selected, these will be shown as comma separated lists of filename.ext.

* Each file (displayed as filename.ext) inside the file input box would be clickable, and individually adjustable by doubleclicking on said file. Clicking the file would select it, and enable the user to remove that file from the upload.

* Dragging one or multiple files to the file input box would add these files to that box.

Having a status bar would be great, but it would also require some additional HTML attributes or even nested FORMS. Not standards compliant.

However, if it’s possible to know for the browser when it’s uploading a file, it’s probably easiest to put every form submit inside a different thread, enabling the user to continue using his current browser window and tab to do whatever he wants. Whenever the browser submit is done, a notice would appear (e.g.: “Your form submission to labs.mozilla.com has been completed. Click here to continue.”)

Clicking that notice would open the destination of that form in the current page. Middle-clicking would open that window in a new tab. Ctrl + clicking would open it in a new window.

It would be a radically different approach to submitting forms, but it would never lock the user down in any possible way.

It would also enable Mozilla (Firefox) to display a statusbar for each form submission in the bottom browser bar, or even in the current tab.

Optionally, if you can handle each file input individually, add a small 2-pixel height status bar to each file input, spanning the entire width of said file input.

The problem often cited against doing things this kind of way is that the browser doesn’t know how much the server has *really* received. But it’s better to have a progress bar that is informative 99.99% of the time than to have no guidance at all 100% of the time. Server-side ACKs can be introduced in a later iteration quite easily.