eHeaders does not
make sense before readyState is 3 or 4, raising a DOMException
INVALID_STATE_ERR makes more sense. That way, these two methods would
be aligned with status and statusText attributes.
There is a related inconsistency with respect to the behavior of responseXML and responseText before readyState is 3 or 4. Currently responseText is required to be an empty string, but again, an INVALID_STATE_ERR makes more sense. I would argue for the same behavior for accessing responseXML before readyState is 4. IE 6 behaves in this manner throwing an exception with message "The data necessary to complete this operation is not yet available."
Subbu
There is a related inconsistency with respect to the behavior of responseXML and responseText before readyState is 3 or 4. Currently responseText is required to be an empty string, but again, an INVALID_STATE_ERR makes more sense. I would argue for the same behavior for accessing responseXML before readyState is 4. IE 6 behaves in this manner throwing an exception with message "The data necessary to complete this operation is not yet available."
Subbu
On 8/10/06, Anne van Kesteren <[EMAIL PROTECTED]> wrote:
On Tue, 20 Jun 2006 13:38:19 +0200, Alexey Proskuryakov
<[EMAIL PROTECTED]> wrote:
> "If no headers of that name were received, then it MUST return the
> empty string."
>
> This matches the behavior of Internet Explorer and Opera 9, but not
> of Firefox 1.5 or Safari 2 (the former returns null, while the latter
> returns undefined).
>
> I don't see any restriction on the semantics of empty vs. missing
> headers in RFC 2616, and indeed, the difference is sometimes
> significant, at least in the case of requests (SOAP, Atom API). Thus,
> the Firefox behavior appears the most reasonable to me. It is
> documented in the idl at
> <http://lxr.mozilla.org/mozilla/source/content/base/public/nsIXMLHttpRequest.idl#148 >.
>
> Can the specification be adjusted to match it at this point?
Yeah, fair enough. It also seems more consistent with other methods.
Now I wonder if it should be null before readyState is 3 or 4 as well...
--
Anne van Kesteren
<http://annevankesteren.nl/>
<http://www.opera.com/>
--
------------------------------
http://www.subbu.org
