[webkit-reviews] review requested: [Bug 29121] XMLHttpRequest::getAllResponseHeaders throws INVALID_STATE_ERR when called in state HEADERS_RECEIVED : [Attachment 39471] Proposed patch. Moved tests to http/tests/xmlhttprequest from fast/xmlhttprequest

[bugzilla-daemon at webkit.org]

Carol Szabo <carol.szabo at nokia.com> has asked  for review:
Bug 29121: XMLHttpRequest::getAllResponseHeaders throws INVALID_STATE_ERR when
called in state HEADERS_RECEIVED
https://bugs.webkit.org/show_bug.cgi?id=29121

Attachment 39471: Proposed patch. Moved tests to http/tests/xmlhttprequest from
fast/xmlhttprequest
https://bugs.webkit.org/attachment.cgi?id=39471&action=review

------- Additional Comments from Carol Szabo <carol.szabo at nokia.com>
Alexei,
The text your are mentioning is intentionally left that way. If the test is run
interactively, that is in a regular browser, not with DumpRenderTree the list
of headers and the the value of the header respectively are still displayed.
Also regardless of the user agent the very next paragraph after the one you
mentioned explains that the header list and the header value respectively are
omitted for DumpRenderTree to prevent the failure of the unattended tests.
I have read the discussion that you mentioned. I believe that my change which
leads to an exception not being thrown, not to extra exceptions being thrown,
increases (a little bit, not very much) the likelihood that a webpage will look
as intended, as opposed to the change discussed in bug 15356, apparently would
create a more restrictive behavior not a more lenient one. The risk with my
change is that pages may be created that work only on Mozilla and WebKit or
only on WebKit which is not that bad.
I know that somebody may rely on an exception being thrown instead of the
header value being returned, but I think that that use case may be discounted.