Web Quality Ideas

[Simone Onofri]

Hi Karl and all QA-IG List,

I read documents "Buy standards compliant Web sites"[1] and "My Web
site is standard! And yours?". A good reading for company and private
that would like to have a site and, vice-versa, for web authors who
like to make a great web. So, there are not other articles from 2002.
There are more ideas to expand the Requirements? Like a guide or
reference? Also considering Unicorn[3] can be a good idea, sounds like
a marketing operation, using a Conformance Logo for Web Quality (like
WAI Conformance logo). This can be useful for tracking sites that uses
correctly W3C technologies, a problem - tracking good sites, explained
by Karl in a past mail to list regarding Google Stats.

For ideas, collaboration and writing, I'm available as a volunteer.
Also for Unicorn is great that funds of Contributing Supporters is
used for It. I'm happy for this!

Thanks for Your time and Best Regards,


[1] http://www.w3.org/QA/2002/07/WebAgency-Requirements
[2] http://www.w3.org/QA/2002/04/Web-Quality
[3] http://qa-dev.w3.org/unicorn/

--
Simone Onofri

-----------------------------------------------------------------------------------------------------------------------

W3C Contributing Supporter
Member of International Webmasters Association / HTML Writers Guild
Member of AICA: Italian Association for Computer and Automated Calculation
IWA Certified Web Professional Candidacy Status in HTML
Microsoft Certified Preinstallation Specialist
Microsoft Certified Sales Specialist
Microsoft Certified Licensing Specialist
Brainbench Certified Internet Professional Web Designer
Brainbench Certified Internet Professional Web Developer

-----------------------------------------------------------------------------------------------------------------------

[Karl Dubost]

Hi Simone,
Thanks for this thoughtful email.


Le 06-10-07 à 01:28, Simone Onofri a écrit :
> I read documents "Buy standards compliant Web sites"[1] and "My Web
> site is standard! And yours?". A good reading for company and private
> that would like to have a site and, vice-versa, for web authors who
> like to make a great web. So, there are not other articles from 2002.

Unfortunately. It seems that the participants on the mailing lists  
www-qa and public-evangelist are more readers than authors. :) Though  
it doesn't mean it should or it could not change.

Everyone who is subscribed is a possible participant of QA IG.

> There are more ideas to expand the Requirements? Like a guide or
> reference?

There is still plenty to do. And if you have specific ideas on  
participating by producing documents you are more than welcome. More  
on that later in this email

> Also considering Unicorn[3] can be a good idea, sounds like
> a marketing operation, using a Conformance Logo for Web Quality (like
> WAI Conformance logo).

There is no WAI Conformance logo. There are WCAG logos for each  
level. Those are attached to a specification. There are a lot of  
issues with regards to the quality which makes difficult to promote  
already a set of simple guidelines.

> This can be useful for tracking sites that uses
> correctly W3C technologies, a problem - tracking good sites, explained
> by Karl in a past mail to list regarding Google Stats.

It is indeed important to use correctly W3C technologies.

  1. We can track Web sites which do a good use of
     W3C Technologies. This kind of lists are usually
     hard to maintain and difficult to define. With
     the risk of false claims, with the risk of
     having outdated information.

  2. We can help people to have a better understanding
     of the W3C Technologies. By giving them tools to
     check and by giving them documentation which is
     suitable for their needs.

The 1. is almost impossible to achieve. We would have to define what  
is a good use of the technologies. There are criteria which are not  
easily checkable because of the specification themselves, because of  
authoring or coding practices, because of the software implementations.

Let's a very simple case a static Web page written in XHTML 1.1 with  
CSS instructions. How do we define the good use for this simple case?  
How do we "certify" the quality? Some of the things to check:

   - Validity of XHTML 1.1
   - Correct use of mimetype application/xhtml+xml
   - Correct use of CSS
   - Correct Mimetype for CSS
   - Language of stylesheets is declared or not if we do not use  
style element or externql CSS but only style attributes.
   - HTTP. Are the http headers used appropriately?
   - What about the semantics of the elements? Are they accordingly  
with what the specification is saying?
   - Do we include WCAG checking?

And then, there is the process, for this only one page, when and how  
do we check that the page respects the defined criteria?

Now, multiply this on a Web site of thousands and/or million of Web  
pages and you get a huge machinery which in a context of Web services  
would be impossible to manage. If one page fails, does the site loses  
its quality label? Which percentage of failure is acceptable?

It doesn't mean that we have to give up on quality, but that it's not  
a label, it's an ongoing process. There are initiatives in these  
directions. For example, Laurent Denis, Eli Sloim have worked on a  
Quality Framework which is very useful to keep track of your quality  
criteria when creating/maintaining a Web site.

> For ideas, collaboration and writing, I'm available as a volunteer.
> Also for Unicorn is great that funds of Contributing Supporters is
> used for It. I'm happy for this!


There are many things which can be written. But more than asking  
someone to write something specific, it's often better to ask the  
person what kind of things they can write. For example,
        - how to sanitize your HTTP installations, a kind of practical  
practices which can be used in the context of CHIPs and Web arch.
        - Collecting best practices in HTML.
        - Writing some technologies 101 to be published on the QA Weblog  
about the technologies which are being developed. Much needed. I  
would do definitely this part if I had more time, but it is open to  
everyone on this list. If there is a technology that is being  
developed and you want to explain it to the Web communities in simple  
terms. PLEASE DO!  Contact Olivier Théreaux or me and we will arrange  
your contribution on the QA Weblog.




--
Karl Dubost - http://www.w3.org/People/karl/
W3C Conformance Manager, QA Activity Lead
   QA Weblog - http://www.w3.org/QA/
      *** Be Strict To Be Cool ***

[Stéphane Deschamps]

<quote who='Karl Dubost' when='10/10/2006 08:55'>

> Let's a very simple case a static Web page written in XHTML 1.1 with CSS
> instructions. How do we define the good use for this simple case? How do
> we "certify" the quality? Some of the things to check:
>
>   - Validity of XHTML 1.1
>   - Correct use of mimetype application/xhtml+xml
>   - Correct use of CSS
>   - Correct Mimetype for CSS
>   - Language of stylesheets is declared or not if we do not use style
> element or externql CSS but only style attributes.
>   - HTTP. Are the http headers used appropriately?
>   - What about the semantics of the elements? Are they accordingly with
> what the specification is saying?
>   - Do we include WCAG checking?
>
> And then, there is the process, for this only one page, when and how do
> we check that the page respects the defined criteria?

And then, there is also the practice itself: two CSS designers have each
their own way of writing their CSS code, who's to say that one is better
than another.

Samely, I'm prone to using several H1's on a page, whereas many people
feel that it's a betrayal of the spirit of HTML (hierarchical structure
seen in a pyramidal mental model) to use more than on H1 per page. Who's
to say when this is wrong or right, or simply best, to use one, two or
three level-one headings?

(these points are made to illustrate the difficulty of certification).

> Now, multiply this on a Web site of thousands and/or million of Web
> pages and you get a huge machinery which in a context of Web services
> would be impossible to manage. If one page fails, does the site loses
> its quality label? Which percentage of failure is acceptable?
>
> It doesn't mean that we have to give up on quality, but that it's not a
> label, it's an ongoing process. There are initiatives in these
> directions. For example, Laurent Denis, Eli Sloim have worked on a
> Quality Framework which is very useful to keep track of your quality
> criteria when creating/maintaining a Web site.

We have the same problem at the moment in my company with applying the
WCAGs. technically, a criteria is not fulfilled if it's not 100% ok. For
instance, you may have 100 images on a page, but we in my company say
the work is not done if one image has no alt text. Should we say it's
99% good, or should we say it fails the criteria?

For lack of time, I can't help more with evangelism and QA, but oh,
would I love to.

--
Stéphane Deschamps
   personal: http://www.nota-bene.org/
   org: http://www.evolt.org/
   french org: http://www.pompage.net/