Re: More on SVG within HTML pages

Jonas Sicking wrote:
> On Tue, Sep 8, 2009 at 7:22 AM, Shelley Powers<shelleyp@burningbird.net> wrote:
>   
>> Thomas Pike wrote:
>>     
>>> Shelley Powers wrote:
>>>
>>>       
>>>> Henri Sivonen wrote:
>>>>
>>>>         
>>>>> On Sep 8, 2009, at 16:14, Shelley Powers wrote:
>>>>>
>>>>>
>>>>>           
>>>>>> Bits left out of the SVG file will also make themselves known, very
>>>>>> quickly.
>>>>>>
>>>>>>             
>>>>> There's existing content out there that contain inexplicably
>>>>> copy-pasted partial SVG in text/html. Presumably this is pure cargo
>>>>> cult and the authors don't even expect vector graphics to appear.
>>>>> However, if browsers wreck the rest of those pages, the user
>>>>> perception would be that the new browser doesn't work.
>>>>>
>>>>> (URLs in Hixie's posts to this list.)
>>>>>
>>>>>
>>>>>           
>>>> I would imagine that a person making this mistake, and then forming a
>>>> judgment that the new browser is broken, will probably find many other
>>>> ways to "break" the browser.
>>>>
>>>>         
>>> The point here is that the person making this mistake (the author) is
>>> not the same entity as the person seeing the problem (the user).
>>>
>>> The author would not notice the mistake because he/she is using a
>>> pre-HTML5 browser.  But all users of a browser that is one of the first
>>> to implement SVG-in-HTML would see the rest of the page as missing.
>>> These users would blame their upgraded browser, and the author of the
>>> page would remain in blissful ignorance of his/her mistake.
>>>
>>>
>>>       
>> If the page does not show correctly, most people send an email to the page
>> author, or ignore it.
>>
>> I doubt that many people will surf the web, come upon a single page that
>> isn't working and make an assumption that the browser is broken. Typically,
>> it requires many such pages for people to make that assumption.
>>
>> We have to operate under the assumption that most people act reasonably. We
>> can't make all of our decisions based on unreasonable behavior. If we do
>> this, we should stop anything to do with CSS, SVG, Canvas, Script, HTML, and
>> so on, because all of it can be used incorrectly, and people act
>> unreasonably or even irrationally because of it.
>>     
>
> I'm unclear on exactly what you are proposing. Is this just a proposal
> to have HTML5 validators ignore any SVG fragments inside the HTML
> page? Or something more, such as change the HTML5 parsing algorithm?
>
> Having validators validate SVG fragments according to the full SVG
> (1.1?) spec makes a lot of sense to me.
>
> / Jonas
>
>   
And as people have said, someone will make a polyglot validator someday, 
but until then, all I'm doing is discussing how to handle namespaced 
elements within SVG.

Right now, there is some confusion about this in the HTML5 spec, and 
right now validator.nu throws out an amazing number of "errors" for 
things that are not errors.

Shelley

Received on Tuesday, 8 September 2009 17:16:19 UTC