The state of our web site and standards

[Daniel Veillard <veillard redhat com>]

  Okay, this is yet another rant from the standard checking department,
and this time there is no binary compatibility issue I assume. Of course
like all good rants there is at least a couple of practical suggestion so
don't trash this mail right now.

  Basically our web site is looking like a assertion that the Gnome
project don't care about standards compliance. Let's take our main
web page and try to check it's conformance:

  http://validator.w3.org/check?uri=www.gnome.org&doctype=Inline

  Result: 12 errors reported basically of 4 types
     - invalid attributes
     - duplicate attributes on an element
     - improper closure of an element
     - absence of URI-Reference escaping for some anchors
       http://www.w3.org/TR/html401/appendix/notes.html#h-B.2.2

  Our news site:
  http://validator.w3.org/check?uri=http%3A%2F%2Fnews.gnome.org%2Fgnome-news%2F&doctype=Inline

  Result: 50+ errors

  Just for the record, http://www.kde.org/ tries to comply to XHTML-1.0
(okay they broke it by adding ISO-Latin-1 character in a document labelled
as UTF8, so they have some work left to do too :-). The fact that they
actually designed a complete browser within the project probably helped, but
this kind of results are not the result of good luck, this proves
something about the project.

  However I'm very happy to see that http://developer.gnome.org/
is a valid HTML page :-).

  What can be done:
   - well fix the home page, and try to fix the generation engine for
     the news site too.
   - make sure when evaluating new designs for the web site (there is
     some competing projects) or new tools for the web site that they
     are clean. For example if we find that X or Y engine for the news
     site makes it impossible to generate valid HTML, eliminate it from
     the list of tools to use.
   - try to automate error reporting. Since at least a large part of the
     site is in CVS it is possible to automate the verification, the 
     engine for the validator used at W3C is available from 
     http://validator.w3.org/source/
     automating the checking from a CVS checkout may not be trivial
     but should be relatively easy. There are others validators available
     in source code.
   - try to automate fixing, there is at least one tool I know available
     in source for doing this, its Tidy:
     http://www.w3.org/People/Raggett/tidy/

  I'm not sure how much people would appreciate an automated solution doing
the checking and commiting tidy'ed up result. But I'm pretty sure it would
convince people to make sure beforehand that their pages are valid :-)

  I still prefer a simpler but valid Web page design than flashy but horribly
broken stuff, I don't think I'm the only one ;-)

Daniel

-- 
Daniel Veillard      | Red Hat Network http://redhat.com/products/network/
veillard redhat com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/