jim.ley at gmail.com
Wed Jun 30 01:30:26 PDT 2004
On Wed, 30 Jun 2004 08:47:19 +0200 (METDST), Martin Kutschker
<martin.t.kutschker at blackbox.net> wrote:
> > As for the date the user inputs, the UA should use the user's locale
> > to determine the timezone.
> What's the users locale? OS/UI settings? Browser settings?
I thought I'd already started putting to bed the myth that the user
has accurate clocks.
Let's look at Opera on Symbian 60 - The OS and the Browser don't have
anywhere to put the timezone in, you can just set the time. Hixie
then told me that the timezone comes from the GSM layer - great,
except of course that relies on the user changing their time everytime
they change time-zones, people just don't do that with their phones,
their watches maybe, but not their phones. So that platform, the
device, the user and the server are all going to have different ideas
on what time it is.
We don't trust the users to send UTC other times (we have no need to,
unless we're asking for a UTC time) there simply aren't the use cases,
or the realisitic deployability of this.
> That's what they currently do. Wasn't the idea to improve this?
That's what currently works, and is understood by all your users,
you're going to need something an awful lot better, and an awful lot
easier to use - remember most users already know 11 tab 04 for their
credit card expiry for example, changing that to widgying around with
a mouse doesn't improve things, the proposed UI needs to be orders of
magnitude better for people to want to use them, given the inherent
problem in all legacy UA's - of which there are a billion or so.
> Or the spec defines that a date should be displayed in the
> shortest possible way (no names of weekdays and months).
I agree knowing the scope of the size of the control is useful, but
names of months are also great for display, everyone can deal with
those who speaks the language of the page.
More information about the whatwg