I just updated my Palm-Linux page, adding specific Sony Clié hotplug script for PPP connections.
It is really useable now!

Next step: The doc is somehow messy, with Palm/Clié and Linux kernels 2.2.X/2.4.X mix. I will separate clearly the two cases, even if this implies some duplication of text.

No change compared to 0+5i version. It ran Out Of the Box, after I made the same changes.

I added a small script, as I really want some files to keep the same date. This script looks for files (in blog directories) which names look like a date, then it changes the date of these files with the name itself.
It runs every hour, and could be used with other versions of blosxom as well.
Here is the source of this file:

#! /bin/sh

for i in `find ./blog -name '*.txt' -print`
do
    DATE=`expr match $i '.*\([0-9]\{12\}\).txt'`
    if [ -z ${DATE} ]
    then
        DATE=`expr match $i '.*\([0-9]\{8\}\).txt'`
        [ ${DATE} ] && "DATE=${DATE}1200"
    fi
    [ ${DATE} ] && touch -t ${DATE} "$i"
done

This script accepts files like "*DATE.txt", where DATE can be in both YYYYMMDDhhmm.txt or YYYYMMDD.txt formats. In the latter case, a default of 12:00 is used for the time.

My version of gmplayer (0.90pre9-3.2) was sometimes very slow on playing mpeg movies (and only mpegs), while mplayer was running fine. The error message was:

         ************************************************
         **** Your system is too SLOW to play this!  ****
         ************************************************
!!! Possible reasons, problems, workarounds: 
[etc...]

The solution was to find the "cache = 4" line in ${HOME}/.mplayer/gui.conf, and change the value from 4 to 512. Everything is perfect now!
Maybe you could experiment with different values of cache, like 1024, or 2048.

Two new big changes, in the path to migrate this site to a fully XHTML/CSS web site: The "Webcam" page will disappear (being replaced by a more useful "GPG" page - already done in all new pages), and the "Palm" page has been rewritten in XHTML.

1. Change the new links (Webcam vs GPG everywhere). Easy. Done on Jan 7.
2. Rewrite the "Weather" page in XHTML. Easy. Done on Jan 11th. But it was not easy at all! I was not able to include a jsp in middle of a SSI file, so I run a hourly job to prepare an html file instead (with an ugly wget getting the data from the jsp). Any better idea on this?
3. Do something about the "Forum" page. I will probably remove it (I don't like their licence sytem - after all, even if it was small help, I crontributed to this software!). Done on Jan 11. Just by removing links, and redirecting the forum page to the main page.
4. Rewrite the "Linux" page. This will be longer: I have 2 ways here. I could just make it XHTML compliant, or make it more accurate. I will probably choose the first solution. Done on Jan 12th.
5. Rewrite the "Gallery" page. Again, I will probably change the engine. Not easy, and I prefered to "patch" the old system to be compliant with XHTML standard. There are problems with images overriding he right column - I'm not going to solve them soon! Done on Feb 9th.

As I just rewrote my "Links" page in xhtml, I wondered if I should include it in the weblog or not.

More generally, my question is: Where should a Weblog stop? Should technical data (such as Blosxom installation) be there?
As I don't have an answer yet, I will leave the Links page as it is, as well as the Forum one (but I think that one will disappear as soon as Blosxom can handle postings :-)

I am using Blosxom (version 0+5i). As I needed a special setup (more than one weblog per page, SSI includes for header/footer, etc...), I could not use Blosxom as a "page generator", but only as an "article manager" inside an outside page.

I made the following to make it work:

• First, in my html index file (server parsed), I included the following:

     <!-- I call blosxom.cgi with the parameters.
          If none, the default "/blog/diary" is used -->
     <!--#set var="BLOSXOM_CATEGORY" value="/blog/diary" -->
     <!--#if expr="$DOCUMENT_PATH_INFO" -->
       <!-- #set  var="BLOSXOM_CATEGORY" value="${DOCUMENT_PATH_INFO}" -->
     <!--#endif -->
     <!--#set var="PROGRAM" value="/cgi/blosxom.cgi${BLOSXOM_CATEGORY}" -->
     <!--#set var="RSS_PROGRAM" value="${PROGRAM}/?flav=rss" -->
  

  Then I use the PROGRAM and RSS_PROGRAM variables as following:

     [...]
     <a href="<!-- #echo var="RSS_PROGRAM" -->">
        <img src="/images/xml.gif" alt="Syndicate this site!" />
     </a>
     [...]
     <!--#include virtual="$PROGRAM" -->
     [...]
  

  The target is to generate the correct Blosxom path, and to have a default "/blog/diary" one if none is specified.
• I also needed to create a .htaccess file in the directory where the index.html resides, to force an http redirection to the index.html if only the directory is specified in the URL. This is necessary, as only URL like http://www.raoult.com/diary/index.html/path/to/weblog will work (and not http://www.raoult.com/diary//path/to/weblog).

     DirectoryIndex redirect.html index.html index.shtml
     Redirect /diary/redirect.html http://www.raoult.com/diary/index.html
  

• Then, I added the following line in blosxom.cgi, to handle possible "//" (slash slash) between the $url & $path macros:

     $url =~ s/^included:/http:/; # Fix for Server Side Includes (SSI)
     my @pi = split m{/}, path_info();
     shift @pi;
  
     ########## br, to skip initial // (SSI pb)
     $pi[0] eq '' and shift @pi;
  
     my $pi_bl = '';
     while ($pi[0] =~ /^[a-zA-Z]\w*$/) { $pi_bl .= '/' . shift @pi; }
  

• As I don't want to use the <span> tag as a date separator, I commented out the corresponding lines:

     # $content_type eq 'text/html' && $curdate ne "$dw, $da $mo $yr" && 
     # print span({-class=>'blosxomDate'}, $curdate = "$dw, $da $mo $yr"); 
  

• Finally, I run a hourly cronjob to generate the weblog index file:

     #/bin/sh
     # find the directory tree and build a list...
     #
     find blog -type d -print | awk '
     BEGIN {
          URL = "http://www.raoult.com/diary/index.html/"
          print "<div class=\"list\">";
          print "<div class=\"listtitle\">";
          print "Weblog tree...";
          print "</div>";
          print "<ul>";
     }
     {
          printf("<li><a href=\"%s%s\">%s</a></li>\n", URL, $0, $0);
     }
     END {
          print "</ul>";
          print "</div>";
     }' > tree.dat
  

  This file is included in the index.html file as:

     <!--#include virtual="/blosxom/tree.dat" -->
  

That's all folks!!

I lost my first place on Google (for a member of French Parliament), and being between this one (Eric Raoult) and a French Senator (Paul Raoult, #3 on Google) makes me somehow uncomfortable ;-)

But I worked a little, and I have got a new home page!
The changes are:

* XHTML 1.0 instead of HTML 4.01
* No more frames!! (using frames is Bad [tm]).
* Only CSS for presentation. Using HTML tables is Very Bad [tm]
* blosxom is the engine for this page (and for RSS syndication)
* blagg is the RSS aggregator (the right column with news)

What I need to do now is to make the change for all the site (this should not be very difficult).

I was noticed today by a colleague that Google gives my site as no 1 when searching raoult. Cool!