Re: Indicating XHR conformance

Le Wed, 10 May 2006 18:30:50 +0300, Mark Nottingham <mnot@yahoo-inc.com> a  
écrit:

> I'm wondering if it would be useful to have some optional flags in  
> implementations that developers can use to test for. While many JS  
> features can be tested for by testing a property, this isn't so for XHR;  
> e.g., you can't easily test to see if it supports more than GET and  
> POST, for example, or if the cache is correctly implemented.
>
> For example, we could specify a tree of properties on the navigator  
> object that are tied to passing the test suite; authors could use that  
> to figure out if certain features are supported. E.g.,
>
> navigator.w3c.xhr.full  // passes the full test suite
> navigator.w3c.xhr.core // passes all of the MUST-level requirements in  
> XHR
> navigator.w3c.xhr.cache // passes all HTTP caching-related tests
> navigator.w3c.xhr.redirect // passes all HTTP redirect-related tests
>
> There are many ways that this could be done, of course; that's just an  
> example.
>
> The biggest downside I can see is that a vendor might not correctly set  
> them (through accident or maliciously). However, even in that case, it's  
> a lot easier for authors to test for the property and then make an  
> exception or two, rather than test every browser and maintain a  
> whitelist in code.

I'm wary of such things. I would have agreed with this, not long time ago.

When developers have to have their sites compatible with more and more  
user agents (on various devices)... such features are just an added layer  
of headaches for developers who use them.

I would never entirely rely on such features, most likely not at all.

That's because I don't trust the user agent: it can provide false  
information on purpose and/or by mistake (bugs).

-- 
http://www.robodesign.ro
ROBO Design - We bring you the future

Received on Wednesday, 10 May 2006 19:31:45 UTC