I have been away for a while and I have not been able to contribute
much to the discussion about the web site. I haven't finished reading
all messages that have been posted since last week but, like Branko
Collin, I am worried that the discussion has focused on technical
issues related to the presentation, and not much on the contents.
Although some of the contributors to these discussions will probably
not agree with me, I think that the Gimp web site should _not_ be
designed in a way that encourages visitors to come back often. The
Gimp users should visit www.gimp.org when they are looking for a
tutorial or when they want to know how to download the latest version,
and so on. But I would like to avoid any web gimmicks that are
encouraging visitors to come back because of the site itself (not
because of the Gimp - the application) such as online polls,
discussion fora and other dynamic contents. Gimp news is very useful,
but that's enough dynamic contents IMHO.
So I am pretty much in favor of a mostly static site, without any
forum or annotation system.
In the company where I work, we have a template system for HTML pages.
But contrary to most systems in which all HTML pages have to include
the header and footer for the template explicitely, we are doing
exactly the contrary: a script loads the HTML files, scans them for
special meta tags, and applies the template to the pages. The meta
tags are optional and define some properties for the page, such as the
name of the author, the revision info, to which logical section of the
site this page belongs to, etc. So the header of a page could look
like this:
<head>
<title>Gimp.org - tutorials</title>
<meta name="gimp.org.section" content="docs">
<meta name="gimp.org.author" content="wilber@xxxxxxxx">
<meta name="gimp.org.navigation" content="leftbar-style3">
...
Since all these tags are optional, a standard HTML page without any
meta tag would still get the default template applied to it. This
makes it very easy for people to edit the pages, since they only have
to edit the contents without having to worry about the layout. Only
those who want to change some properties of the template will have to
insert these additional meta tags. The files that are checked in CVS
are the "source files", without the templates (of course the templates
are also in CVS, but in a separate module).
In order to discourage people from editing the generated pages (the
ones that are sent by the web server, including the template stuff),
all pages that are generated by the script start with a comment
similar to this:
<!-- This page was generated automatically from xxxxx -->
<!-- For more information about how to edit this page, read zzzzz -->
Note that I also maintain some web sites based on PHP and some others
using custom Perl modules for generating the pages. But in the long
run, I have found that the sites that are kept up-to-date for the
longest time are the ones in which all the non-technical contributors
can simply store their simple HTML files in the CVS repository and
have the template added for them, without having to deal with PHP,
Perl or any special things to include in the HTML file (all meta tags
are optional). By the way, the CVS commit script processes the page
through HTML Tidy in order to ensure that the HTML code is correct.
Nobody can commit a page if Tidy considers it broken.
Yes, this is sometimes frustrating for those who want the latest and
greatest HTML layout for their pages, but in the end this is better
because anybody can maintain any page without much effort.
-Raphael
P.S.: I am posting this to both lists (gimp-developer and gimp-web)
because I suppose that many people did not have the time to
subscribe to the gimp-web list yet. But if you reply, consider
sending your message to the gimp-web list only.
