Re: 9. WYSIWYG editor (enforcing the signature) detailed review of

Hi Mihai,


On Aug 3, 2007, at 4:46 AM, Mihai Sucan wrote:

>
> Hello!
>
> I have read the HTML 5 spec section on WYSIWYG editors [1] and I'd  
> like to
> express my concern on requiring the inclusion of "(WYSIWYG editor)"  
> in the
> META NAME="generator" CONTENT attribute value.
>
> Having experience with working on my own WYSIWYG editor, I believe the
> aformentioned requirement is not really applicable. As in, I  
> wouldn't like
> to include the suggested string. Here's a simple reason: my CMS  
> generates
> correct, semantical markup (or tries to do so). The *strict*  
> content of
> the pages is all generated by my WYSIWYG editor, found in the CMS.  
> Having
> the "(WYSIWYG editor)" mention in the "generator" meta-tag does not  
> really
> reflect the reality, when someone looks into the page code, either
> personally, or with a parser. For example, the blog pages are mostly
> generated by the blog engine, and only the article itself is edited  
> with
> the WYSIWYG tool. The user never edits *the entire* page with the  
> WYSIWYG
> editor. The page is not generated by the tool, it's not the result  
> of any
> "silly hacks" in the WYSIWYG world. Obviously, I consider the  
> signature
> appropriate for documents entirely generated by WYSIWYG editors,  
> like NVU,
> Dreamweaver and such.
>
> Another point for not requiring the inclusion of "(WYSIWYG editor)"  
> in the
> meta-tags is: someone will find this "delightful" and just "great" for
> some of his/her parsers, to be able to detect pages generated by  
> editors.
> No means are provided to detect that HTML 5 is used in the document (a
> version attribute, a DOCTYPE which specifically tells the version  
> somehow,
> or whatever), but you provide something which allows parsers to  
> sniff the
> document for "WYSIWYG editors". This is like checking for a  
> "feature", a
> characteristic of the page. Also, this can be used just like the  
> current
> <!DOCTYPE html> to check if the document is "HTML 5" - nobody should
> actually do it, but some will definitely do it. Of course, doing  
> this *is*
> wrong: you cannot rely on the presence of the "(WYSIWYG editor)"
> signature, you cannot rely on the DOCTYPE, you cannot rely on ...  
> anything.
>
> The spec wording even seems ambiguous in the definition of the FONT
> element. Is it allowed on pages without the signature? Is it not? I
> believe it is not allowed because the spec says the FONT element  
> "must not
> be used except by WYSIWYG editors". This causes the following in a
> validator scenario:
>
> 1. if the document does not contain the signature, and no FONT, the
> document is valid. (assuming the document is otherwise valid)
>
> 2. if the document does not contain the signature, but it does use the
> FONT element: the document is invalid - error found "the page seems  
> to be
> generated by WYSIWYG editor, because it uses the FONT tag. You MUST
> include the (WYSIWYG editor) signature in your meta-tag generator  
> content
> value" (with a better wording than mine :) ).
>
> 3. if the document does contain the signature, and no FONT, the  
> document
> is valid.
>
> 4. if the document does contain the signature, and it uses the FONT
> element: the document is valid. According to the spec, this is not an
> error -the use of the FONT element is allowed in this case. Maybe the
> validator should warn on the use of the FONT tag ("try to improve the
> semantic of your code"), but that's another story.
>
> I don't "like" the above scenario. Having the document as valid,  
> is ...
> practically... at the "mercy" of the generator meta-tag. I do not  
> consider
> this anywhere close to "appropriate".
>
> I suggest that the editors of the spec remove this requirement,  
> because it
> simply adds another rule to be broken/missused. Thus, the FONT element
> shall be allowed on any page, irrespective of the generator signature.
>
> If this request does not seem convincing, please provide your  
> arguments
> for keeping the spec the same. What are the use cases? Any examples  
> where
> this signature is needed today? Which (types of) UA would need the
> signature?
>
> [1] http://www.w3.org/html/wg/html5/#wysiwyg

I'm past due to submit my review of this section, but I'm glad to see  
someone has done it too. I agree with the point you make about the  
meta signature. This is another one of those areas where the use- 
case / problem statement is not clear to me at all. Often times I can  
glean something regarding the use-case from reading the solution.  
However, in this case I can't.

One related thing I think might be useful, might be to encourage all  
editors (WYSIWYG or not) to include a editor user agent string just  
as browsers do. In this way we could have some indication of how some  
code got generated. If an author is creating everything by hand, then  
the UA string would be absent. However, in any case of an editor that  
assist an author in creating content it would be expected to add it's  
UA string.

The question in either of these solutions (again, I'm not sure the  
problem so they might not be solving it), is what should happen when  
an author edits a document in another editor, and another, and  
another? Should every editor ensure it's UA string is there? If we're  
trying to thoroughly trace the editor interaction with a document,  
then we would need that. Not sure if that's what we want though.

If anyone can state the use-case / problem statement on this editor  
signature, that would be very helpful.

Take care,
Rob

Received on Friday, 3 August 2007 13:30:29 UTC