[webkit-changes] [24935] trunk

[Alexey Proskuryakov]

Hi Kevin,

  You are right that HTML 5 only specifies the behavior of createElement for
HTML documents (which is not directly related to HTML serialization though).
However, the concept of "a Document created with an XHTML MIME type" is
completely alien to this spec AFAIK, so our current behavior is in rather
bad contradiction with it.

  Also, Hixie's comments in bug 8007 and other ones seem to indicate that
the intention is to make namespace-unaware methods behave consistently in
XML and HTML documents as much as possible.

  Looking at Zimbra's bugzilla, I believe that yes, their TOT works with
Safari 3 (pre-r24935). I haven't tested this myself, though.

on 13.08.2007 22:59, Kevin McCullough at kmccullough at apple.com wrote:

> Hi Alexey, hopefully I am explaining this correctly.  r24935 was aimed
> at changing crateElement wrt XML serialization (as opposed to HTML
> serialization) which I don't believe is specified in the HTML5 spec,
> and the change is purely to match compatibility with FireFox (the
> differences I saw I turned into the layouttests).  Are you sure Zimbra
> workes with TOT before r24935?
> 
> If I'm unclear on something please feel free to clarify
> 
> • Kevin McCullough
> kmccullough at apple.com
> 
> 
> 
> On Aug 11, 2007, at 10:36 PM, Alexey Proskuryakov wrote:
> 
>> 
>>  So, Zimbra already has code that works with TOT WebKit, and doesn't
>> need
>> any workarounds. Do we have any other reasons to be non-conformant
>> with
>> HTML5? If not, I'd suggest reverting r24935.
>> 
>> on 09.08.2007 21:49, Kevin McCullough at kmccullough at apple.com wrote:
>> 
>>> For that decision I will forward you on to Hyatt and Maciej.
>>> 
>>> • Kevin McCullough
>>> kmccullough at apple.com
>>> 
>>> 
>>> 
>>> On Aug 8, 2007, at 9:53 PM, Alexey Proskuryakov wrote:
>>> 
>>>> Hi Kevin,
>>>> 
>>>> Thank you for the clarification!
>>>> 
>>>> Do you expect that this will make it to HTML5, or will WebKit be
>>>> non-conformant from the very beginning? HTML5 seems to have a nice
>>>> model
>>>> (old-style DOM functions always work in HTML namespace), and doesn't
>>>> have a
>>>> notion of "a Document created with an XHTML MIME type" AFAICT. A
>>>> real life
>>>> site that breaks on an "isSafari" branch is not a very convincing
>>>> example.
>>>> 
>>>> on 09.08.2007 02:16, Kevin McCullough at kmccullough at apple.com
>>>> wrote:
>>>> 
>>>>> Hi Alexey,
>>>>>    Sorry I wasn't clearer.  I reverted r24146 and added some code
>>>>> so
>>>>> now you will get an XHTML namespace when calling CreateElement if
>>>>> you
>>>>> are an HTMLDocument or a Document created with an XHTML MIME type.
>>>>> 
>>>>> I will update the bugs below with this information.  I think 14835
>>>>> can
>>>>> probably be closed, and the rest just need a comment.  This does
>>>>> fix
>>>>> the given error in 11565 but we are still unable to log into Zimbra
>>>>> for another reason.
>>>>> 
>>>>> I think 8007 is where the conversation for this issue should take
>>>>> place.  I'll add a comment.
>>>>> 
>>>>> • Kevin McCullough
>>>>> kmccullough at apple.com
>>>>> 
>>>>> 
>>>>> 
>>>>> On Aug 8, 2007, at 1:22 PM, Alexey Proskuryakov wrote:
>>>>> 
>>>>>> on 08.08.2007 21:14, kmccullo at webkit.org at kmccullo at webkit.org
>>>>>> wrote:
>>>>>> 
>>>>>>>     - <rdar://problem/4976879> REGRESSION: Safari doesn't work
>>>>>>> with Zimbra
>>>>>>> enhanced login.
>>>>>>>      - Reverting a previous change, and modifying how documents
>>>>>>> are created
>>>>>>> so that we better match other browsers behavior with respect to
>>>>>>> namespaceURIs.
>>>>>> 
>>>>>> What is the new behavior, and is it documented anywhere? It's not
>>>>>> even
>>>>>> quite obvious which previous change was reverted.
>>>>>> 
>>>>>> See bugs 8007, 14506, 10932, 14835 - I think some or all of them
>>>>>> may need
>>>>>> to be updated to describe the new behavior.
>>>>>> 
>>>>>> +    log("This test was written to address rdar://problem/4976879.
>>>>>> And
>>>>>> tests how namespaceURI's are assigned.");
>>>>>> 
>>>>>> Would it be possible to have this bug reflected in Bugzilla, so
>>>>>> that the
>>>>>> rationale would be known? It's good to know the title of the bug,
>>>>>> but maybe
>>>>>> there was some useful information in the report and in subsequent
>>>>>> discussions?
>>>>>> 
>>>>>> In particular, I'm wondering if this is in fact a workaround for
>>>>>> some
>>>>>> broken Safari-specific code in Zimbra (see bug 11565, which is
>>>>>> probably
>>>>>> different, but may be tangentially related).
>>>>>> 
>>>>>> - WBR, Alexey Proskuryakov.
>>>>>> 
>>>>>> 
>>>>>> _______________________________________________
>>>>>> webkit-changes mailing list
>>>>>> webkit-changes at lists.webkit.org
>>>>>> http://lists.webkit.org/mailman/listinfo/webkit-changes
>>>>> 
>>>> 
>>>> - WBR, Alexey Proskuryakov.
>>>> 
>>>> 
>>> 
>> 
>> - WBR, Alexey Proskuryakov.
>> 
>> 
> 

- WBR, Alexey Proskuryakov.