Re: XML input control

> Do I understand you correctly, that by "rich content" you primarily
> mean textual content? I can also imagine some other uses like SVG,
> but that seems a bit too futuristic.

I agree there are some interesting extensions here, but for the time  
being, lets limit the discussion to content that ordinary users put  
into CMS systems: formatted text, links, (pre uploaded) images,  
paragraphs, lists etc.

> How can this be forced to any client software? There is no way
> you can trust some remote input being valid. You clearly also
> need that XML validator on the server side.

If a client claims to support this feature I think it will be  
reasonable trust the validity of the data returned - just like we  
trust other tags and messages from the clients. Of course this scheme  
will only work when a critical mass of major vendors choose to  
support it.

> And what should you
> do, when you detect an error in input? Do you tell the user,
> who wrote the document with WYSIWYG editor, that there is parse
> error on line 127?

Yes, I think that will suffice. And the user should be advised to  
report the error to the maker of the editor. The whole point of this  
mechanism is to delegate error handling to the frontmost input tool,  
preferably in the form of error *prevention*. If the output of an  
editor does not validate, this should be considered a serious fault  
of that input tool, not something that the user should deal with  
herself.

> The other problem is your suggested fallback to pure XML.
> Well, if we are talking about text, then I would clearly prefer
> a fallback to plain old human-friendly text.

OK. For some types of content, ascii based syntax (e.g. wiki syntax)   
could be offered as a fallback. In addition to pure xml, users should  
be offerd a generic schema driven markup editor as fallback when  
their input tools fail.

> I completely agree, that there is a clear need for a general method
> to enter formatted text into webpage, but it's a really-really
> complex problem.

I think this complexity should be handled by the authoring tools - as  
close to the user as possible. I also think we can make things easier  
by creating limited tailormade document types for these tools. The  
xml control will not solve this problem once and for all, but I think  
it will facilitate development of domain specific solutions over time.

When I get time, I think I will write make a couple of test cases.

--
Henrik

Received on Monday, 26 March 2007 15:10:35 UTC