W3C home > Mailing lists > Public > public-webapps@w3.org > July to September 2009

Feedback on the Progress Events spec

From: Ian Hickson <ian@hixie.ch>
Date: Mon, 17 Aug 2009 22:02:04 +0000 (UTC)
To: public-webapps@w3.org
Message-ID: <Pine.LNX.4.62.0908171937360.31215@hixie.dreamhostps.com>

Is there a record anywhere of outstanding feedback on the progress events 
spec? I looked in the issue tracker but there was only one issue there.

I was trying to integrate HTML5 with Progress Events today and found the 
following problems:

 - ProgressEvent.total has a requirement that I don't fully understand:

    "Where the size of the transfer is for some reason unknown, the value 
    of this attribute must be zero."

   It's not clear if that's a requirement on other specifications, 
   on authors, or on implementations. If the latter, I don't understand 
   how to apply it when the UA is unaware of the relevant operation (e.g. 
   when the event is constructed from script).

 - There's no conformance requirements saying what the attributes on the 
   ProgressEvent interface must return (in particular, I would have 
   expected them to just return the values passed in the initialiser, 
   but it's not clear.) (Note: HTML5 gets this wrong currently on most of 
   its event interfaces also.)

 - The definition of "Firing a Progress event" is not useful for other 
   specs because it contains no normative conformance criteria.

 - In the HTML example, the "irrelevant" attribute got renamed to "hidden" 
   in HTML5.

 - I'm still confused by the duplicate and redundant conformance critiera 
   when they are present. For example, it's not clear to me what the 
   "must"s and "may"s really mean in the table in 2.1 Event definitions; 
   the requirement "User agents must ensure that these events trigger 
   event listeners" seems like it is stepping on the DOM Events spec's 
   bailiwick, "User Agents must implement the events as defined by any 
   such specification" seems like a tautology, etc. I'm very uncomfortable 
   with redundancy in testable requirements, as it makes it quite 
   ambiguous exactly how to test them and what is being tested.

 - The draft appears to be missing a security section (for example, there 
   doesn't seem to be a mention of the possible cross-site attacks 
   possible if the events leak the number of bytes downloaded, 
   especially in the 'error' event).

Modulo those issues, I've now hooked the progress events feature into the 
HTML5 media elements feature.

HTH,
-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Monday, 17 August 2009 22:02:45 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:49:33 GMT