On Thu, 18 Oct 2012, Mikey Clarke wrote:
I'd like a little information on the motivation for using absolute URLs
on input type=url validation.
Currently input type=url is to be validated using absolute URLs.
Thus, 'http://www.mysite.com' validates but 'www.mysite.com' does not. I
consider this to be a huge usability issue. An ordinary user when asked
to provide a URL will be very unlikely to provide the protocol. To an
ordinary user 'www.mysite.com' is the URL, not 'http://www.mysite.com'.
Since most browsers that support both the new input types and that have
fully implemented form validation block submission of a form with
invalid inputs, a user entering 'www.mysite.com' is unable to submit
their form and is instead given an error. Even assuming that the error
notice is descriptive enough to alert to the absence of the required
protocol (this is currently _not_ the case), the user has already been
disrupted. Such strict validation is hostile and potentially confusing
to users. As a developer I currently feel compelled to use the
'novalidate' attribute on forms containing type=url to protect my users
from this behaviour.
I feel that if a developer requires the protocol, they are perfectly
capable of asking the user for it, and doing so in a much clearer way
than the browser itself. If the validation for URL fields is to remain
so strict, I really see little point in this input type being validated
at all; as a developer there is absolutely no way I can use this
validation as it stands, the potential for a poor user experience is
just too evident.
The spec expects browsers to convert www.example.com to
http://www.example.com/; automatically so that this kind of issue
doesn't occur.
--
Ian Hickson U+1047E)\._.,--,'``.fL
http://ln.hixie.ch/ U+263A/, _.. \ _\ ;`._ ,.
Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'