- From: Malcolm Rowe <malcolm-what@farside.org.uk>
- Date: Tue, 15 Jun 2004 16:30:47 +0100
Jim Ley writes: > [select-editable implemented as a fieldset snipped] It sounds like there are still some issues to work out with that solution (which isn't surprising). Could you try to firm it up as a more concrete proposal? It's an interesting idea. > Over client-side validation though is one of the worst problems on the > web, for example email addresses and telephone no's., loads of people > reject my email address in their forms, and having to replicate all > the validation on the server anyway, I'm not sure what the real value > is in the extra validation. The benefit is twofold: immediate feedback, and richer controls. If we can declaratively mark as field as numeric, the UA can prevent the user from entering non-numeric data. It can also choose to present a control specific to numeric entry, maybe a spin-control or similar. I agree that over-validation is very bad, however, Ian's current spec for the 'email' and 'tel' types refers to the relevant normative RFC's, so I can't see any problems there, unless those specs are incorrect (could be true for the 'tel' type, I suppose, highly unlikely for 'email'). It's always possible that a UA might implement it's own ideas about what forms a valid email address, but that's a conformance problem in the UA, not a problem with the spec per se. I'm interested - why would anyone think your email address was invalid? I can't say I've had any problems, though some servers have a tendency to force the whole address to upper case, which is very annoying (and illegal). Regards, Malcolm
Received on Tuesday, 15 June 2004 08:30:47 UTC