W3C home > Mailing lists > Public > public-forms-tf@w3.org > April 2008

Architectural Consistency - What does it mean?

From: Maciej Stachowiak <mjs@apple.com>
Date: Wed, 2 Apr 2008 21:18:44 -0700
Message-Id: <E2C0990F-C815-40B4-A409-16E552820F99@apple.com>
To: public-forms-tf@w3.org


In the interests of making a positive contribution to the Task Force:

"Architectural Consistency" is a pretty broad term. One thing we  
should decide as a Task Force is what sense we intend it in.

Here are some possible ways of interpreting "architectural  
consistency" between multiple forms technologies:

1) Both are consistent with the Web architecture as a whole (in other  
words, URIs for addressing, documents described as markup, REST  
architecture model, etc).

2) Both may be used together on the same Web site without conflict.

3) Both may be used together in the same Web document without conflict  
(for example, through use of XML namespaces to disambiguate).

4) Both are reasonably aligned in their capabilities where they  
overlap, without gratuitous differences.

5) One may be implemented in terms of the other through a prior server  
side translation (this would be a scenario such as "author in XForms,  
translate to HTML Forms for client-side deployment").

6) One may be implemented in terms of the other through client-side  
script-based support (for example, XForms-like markup is sent to the  
client along with a script that translates the mechanisms to HTML  
Forms and implements the processing model).

7) Both must be describable in terms of a single server-side  
processing model.

8) Both must be describable in terms of a single client-side  
processing model.


I would argue that 1-7 are all reasonable expectations for  
architectural consistency. As an example, SVG and HTML would satisfy  
criteria 1-3 and 5-7, and 4 is debatable (there is some overlap in  
areas with differences but it is in dispute whether this is necessary  
or not, and the groups are working on closer alignment).

I would argue that #8 is too strong a requirement. For example, CSS  
and SVG have completely different models for layout. But because there  
are defined ways to interoperate, it is not generally argued that this  
makes them architecturally inconsistent. Similarly, http and ftp are  
completely different protocols from the client's perspective. But  
shared URI addressing and the request-response model bring them into  
an architecturally consistent whole.


Any thoughts from other Forms TF members? Are there other criteria  
that you would see as part of "architectural consistency"? Mine are  
all pretty general to the Web and not very specific to Forms.

Regards,
Maciej
Received on Thursday, 3 April 2008 04:19:33 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 3 April 2008 04:19:33 GMT