Bug#98811: about XHTML compliance

[barbier@linuxfr.org (Denis Barbier)]

On Sat, Dec 15, 2001 at 09:47:53PM +0100, Josip Rodin wrote:
> Hi,
> 
> It would be good if someone provided a thorough list of things we'd have to
> change. It's not like me and other fifty or so people who edit the web pages
> can't go to w3.org and read the 46 pages (+ the DTDs and whatnot) of the
> specification ;) but if at least an overview of required changes was
> available (maybe it exists already?) that would speed up closing of this
> bug.

Hi Josip,

nearly all required changes were already described in the bugreport:
   a. Put all tags and attributes (but not attribute values) in lowercase
      letters
   b. Optional end tags are no more supported, so
           <p>some text here
      must be written
           <p>some text here</p>
   c. Empty tags must contain a trailing slash, e.g.
           <img src="logo.gif" alt="Logo">
      becomes
           <img src="logo.gif" alt="Logo" />
      (extra space is recommended in order to give old browser a chance
      to parse this tag without trying to interpret this trailing slash)
And I would also add
   d. Always enclose attributes within quotes

But there are practical problems, of course ;)
Most of them are due to the way this new trailing slash is handled by mp4h
(WML pass 2).
In short, we could
    1) Apply above changes to templates (this should have no influence on
       generated pages).
    2) Find error-prone constructs, as URLs in consultants/consultant.data
       which may have a trailing slash and then conflict with (b).
       Such errors can be found with
              grep -rl '/>' webwml
       and an easy fix is to add quotes.
    3) Apply (a)-(d) changes to pages (there still should be no changes in
       output).
    4) Manually check that webpages do build fine with
              wml -W2,--expansion=0
       This flag modifies how mp4h parses input, and is suitable for XHTML
       pages.  Since all tags are then considered as being complex, processing
       is broken if trailing slash is missing in empty elements.
       Fix broken pages.
    5) Set this flag in Makefiles to make it the default, and change DOCTYPE
       to XHTML.

I occasionally did (1) as explained in
   http://lists.debian.org/debian-www/2001/debian-www-200108/msg00103.html

Denis