Re: ISSUE-17: Widgets (not just widget engines) should be able to specify which proxies they communicate through

Hi Bryan,

On Thu, Sep 4, 2008 at 6:22 AM, Sullivan, Bryan <BS3131@att.com> wrote:

>
> To remain silent on the proxy configuration for Widgets raises a serious
> problem with interoperability. The rationale above describes why. If a
> particular service environment does not require a proxy, then the host
> device will likely not be configured to use one. But the existence of that
> case is not a valid reason to ignore the case in which a proxy *is* required
> and the host device is configured to use it. In that case, the widget
> framework must use the proxy by default, or it is likely that web
> applications will not work.
>
> Putting the burden on the widget developer to "use a proxy if you need one"
> will not suffice either, because:
>
> -       Widget developers will not always know the specific proxy
> configuration required for each device, and each network which the device
> can access. Having developers hard code or privately configure proxy
> settings is not a scalable practice.
>
> -       Proxy settings do change for various reasons, and are remotely
> manageable (e.g. through OMA Device Management in mobile devices).
> Hard-coded and privately configured proxy settings would be unaffected by
> these changes.
>
> I request that the section on use of the user-configured and default proxy
> be restored.

I understand the problem and I personally don't mind putting it back
in. However, I don't think this requirement can be addressed by
normative text in any of our specifications. That is the reason we
took it out in the first place. I mean, what would the normative text
look like and into which specification of the Widget Family of
Specifications would that text be included?... and how could this be
tested in a test suite?

Kind regards,
Marcos

-- 
Marcos Caceres
http://datadriven.com.au

Received on Friday, 5 September 2008 00:00:46 UTC