On Thu, Sep 11, 2008 at 10:02 AM, Bruce Johnson <[EMAIL PROTECTED]> wrote:
> AFAIK, no one is proposing that <set-property> changes the set of possible
> values. I would certainly argue against that.
> The upshot of all this is:
>
> <set-property> gets another attribute called "values" that is mutually
> exclusive with "value":
>
> <set-property name="foo" value="a"/> // set property 'foo' to
> the value 'a'
> OR
> <set-property name="foo" values="a, b, c"/> // set property 'foo' the
> multi-value {'a', 'b', 'c'}
>
> In every other way, <set-property> has the same semantics as before (e.g.
> must refer to known property values, last one wins, ...).
>
So you would not be allowed to remove an existing property value, and you
could not add a new value in set-property without also calling
extend-property, right?
I think it might be useful to allow removing from the set of legal property
values, and I think having to list all but the one you want to remove is
fragile -- if someone used this feature and then we added, for example, an
ie8 user.agent value, they would silently be excluding the new user agent
and wondering why it wasn't working for them.
But those are relatively minor objections and can be addressed later with a
new tag or extensions to the interpretation of the set-property tag (ie,
<set-property name="user.agent" values="-gecko"/>).
--
John A. Tamplin
Software Engineer (GWT), Google
--~--~---------~--~----~------------~-------~--~----~
http://groups.google.com/group/Google-Web-Toolkit-Contributors
-~----------~----~----~----~------~----~------~--~---
