Re: ISSUE-76: Need feedback on splitting Microdata into separate

Ian Hickson, Thu, 10 Dec 2009 19:23:06 +0000 (UTC):
> On Thu, 10 Dec 2009, Leif Halvard Silli wrote:

>> I don't bet that I am able to impact your conclusion.
> All you would have to do to impact my conclusion is back up your original 
> statement.

(More concretely) on Conway's law possible effects due to a split and 
on HTML 5/microdata itself:

 The  issue in the counter proposal is not microdata, but the 
gluing/communication between HTML and microdata. Conducting the design 
in two documents would supposedly cause the gluing to become: [1]  

 ]]more vulnerable to incompatible changes in the larger spec[[

The counter proposal thus seeks to not destroy our current, ideal (I 
presume) communication structure. However, even if we keep microdata 
inside the HTML 5 spec, still: [1] 

 ]]Reusing [microdata] in, for example, SVG would not be possible[[

I assume that by SVG, Tab also have in mind SVG and MathML sections in 
HTML 5 documents. So how can our communication structure be so ideal, 
when the a lack of glue between _HTML 5 documents_ and microdata has 
developed exactly because of the WHATwg's focus/emphasis on a native 
language only for HTML? 

This way - voila - much of the benefit from microdata's supposedly 
great integration with HTML is gone, as one would would need to use one 
meta data system for the foreign content sections, and another system 
for the HTML sections. 
This outcome is easy to link to our communication structure situation, 
as microdata is so strongly linked to things that went on outside the 
HTMLwg. I believe that a joint effort - a single channel discussion - 
of the entire working group would not have lead to this result. Though 
of course, who knows. It is just a theory about why microdata looks as 
it does. Just as Tab's hint is also just a theory. We are also not done 
yet - who knows what the result finally becomes.

According to Wikipedia [2], Conway's law says that "A" and "B"  will, 
in the end, not "interface correctly" unless they are designed within 
the same communication channels.

 Well, does this mean that you and Tab want us to see the HTML 5 draft 
itself - the specification text - as a communication channel? If yes, 
then I am able to admit that, when looking at it from your side of the 
"keep microdata in the HTML 5 spec" scenario then there is, after all, 
an "A" (HTML) and a "B" (microdata) in it.  However, from where I 
stand, where the HTML 5 draft is not a communication structure, then I 
perceive Tab's counter proposal to say that merely _having_ a "B" is a 
risk. The same impression is created when you insist that microdata 
_is_ HTML and _must_ be part of HTML etc. Conway's law doesn't support 
this, however. The very existence of an "A" and a "B" is not in itself 
defined as a vulnerability by Conway's law, AFAIU. 

But otherwise, if you insist that it isn't possible to see any negative 
interface design effects from our use of different communication 
channels, then I question why cutting out microdata and placing it in a 
separate draft - within the same split communication structure - should 
increase the risks in any serious way. I don't find such a claim to be 
very credible, as the communication channel divide caused by our two 
mailinglists etc, is much more serious than such a document split would 

[ snip ]
> At this point I'm seriously wondering if you in fact just made up your 
> original point with absolutely no basis in reality.  
[ snip ]

At least you did save on the ammunition. What I said was indeed based 
on an application of that "law" (which did not mention any exceptions 
when it does not apply) in combination with my knowledge about our 
communication structure situation, and not on a concrete fact. If the 
law applies, then it applies already. Positively or negatively. Btw, 
feel free to outline the concrete positive effects that have taken 
place due to our split communication structure.


leif halvard silli

Received on Friday, 11 December 2009 02:36:53 UTC