- From: Ian Hickson <ian@hixie.ch>
- Date: Tue, 15 Aug 2006 10:29:37 +0000 (UTC)
On Fri, 26 May 2006, Henri Sivonen wrote: > On May 26, 2006, at 09:23, Matthew Raymond wrote: > > > Henri Sivonen wrote: > > > The WF2 spec does not say anything about filtering autocomplete lists > > > on pattern and Opera 9 does not do it. However, autocomplete entries > > > that would fail form validation are not particularly useful. > > > > > > I suggest saying that potential autocomplete items (whether supplied > > > by the UA itself or read from a datalist) should be tested against > > > the datatype and the pattern of the form field being filled and > > > should only be shown if they match both the datatype and the pattern. > > > > This assumes that the inclusion of such items is unintentional. > > When the values are supplied by the UA (e.g. IRIs from the browsing > history), the inclusion of non-matching items certainly is unintentional > from the author's point of view. Agreed. > > What if you wanted the user to be able to select from a fixed set of > > values that fall outside of the pattern in addition to those that fit > > the pattern? > > Then you'd need to change the form validation phase in the spec. Yeah, this right now wouldn't work anyway. > > Is the proper way to handle this to force web authors to write these > > values into the pattern? > > Yes. (Of course, alternatively, they can remove the pattern and > constrain the field by scripting.) Yup. -- Ian Hickson U+1047E )\._.,--....,'``. fL http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Tuesday, 15 August 2006 03:29:37 UTC