[Bug 20322] Document the force preflight flag

https://www.w3.org/Bugs/Public/show_bug.cgi?id=20322

--- Comment #13 from Hallvord R. M. Steen <hsteen@mozilla.com> ---
So.. tried to write a test case for this to see how current browsers behave.
This should be a valid test:
https://github.com/w3c/web-platform-tests/commit/5fd7fd9b759d12006d2ecba5e26d0026521343d6

Story right now is that:

* Firefox implements what Anne and (apparently) the spec says is correct: don't
fire upload events for "simple" requests, fire for "not simple" requests
(preflighted).

* Opera (Presto) fires all the events, whether the request was "preflighted" or
not.

* Chromium fires none of those events, again regardless of preflight status.

(Can't test in a modern IE right now - it would be a pretty sweet mess if they
did the opposite of Firefox - so we had all possible behaviours covered among
the main four engines..)

Note: the data being sent in the above test case is only 1000 characters -
well, 999 to be precise - maybe sending a more complex payload might change the
way some implementations behave? Thinking of Chrome in particular. I tried to
send a larger payload but at some point it seemed to hit some configured limit
for the python serve.py script..

IMO this mess indicates that this specific corner of the spec isn't relied upon
by authors yet, and we could do a bit of cleanup. I'm not sure how exactly it
should work though - given that it's too late to make sure a preflight request
happens after send() is called, maybe Chromium's "ignore events registered
after send()" model is best?

-- 
You are receiving this mail because:
You are the QA Contact for the bug.

Received on Monday, 25 November 2013 16:12:48 UTC