Ayant une dedibox originale (processeur Via C7) qui était devenue poussive, je me suis dit qu’il était temps de migrer vers une configuration un peu plus costaud.

“Buggar!!” comme dirait un ami à moi: plus de Mandriva sur les nouvelles Dedibox!

Me voilà donc parti pour plusieurs jours de galère pour essayer de migrer mon ancienne configuration. Ayant réussi à démarrer sur un kernel Mandriva, voici la méthode que j’ai utilisée. Un grand merci à Antoine Ginies pour son aide.

Attention: réservé aux utilisateurs Linux avertis! Utilisez cette procédure à vos risques et périls!

1. Installer Fedora sur la nouvelle Dedibox
2. Passer la dedibox en système de secours grâce à la console de gestion noter l’identifiant (ID) le mot de passe et l’adresse (IP)
3. Utiliser le script suivant pour monter le disque dur et passer en chroot

   ssh ID@IP
   sudo mount /dev/sda1 /mnt/
   cd /mnt
   sudo mount -o bind /dev /mnt/dev
   sudo mount -t proc none /mnt/proc
   sudo chroot /mnt /bin/bash
   su - root

4. À partir d’une machine sous Mandriva transférer le système vers la nouvelle Dedibox:

   tar cf - bin/  boot/   etc/  home/  initrd/  lib/   media/  mnt/  opt/   root/  sbin/   tmp/  usr/  var/ | ssh ID@IP "dd of=/mnt/mandriva.tar"

5. Supprimer ensuite la Fedora pour la remplacer par la Mandriva (depuis le chroot de la nouvelle dedibox):

   rm -rf selinux/ bin/  boot/   etc/  home/  initrd/  lib/   media/  mnt/  opt/   root/  sbin/   tmp/  usr/  var/
   tar xf mandriva.tar

6. Voilà la partie la plus délicate: il s’agit de regénérer un initrd qui fonctionne sur la nouvelle dedibox. J’ai utilisé cette commande, mais cela peut varier selon le noyau que vous utilisez:

   mkinitrd --with=ext3 /boot/initrd-2.6.29.2-tmb-server-3mdv.img 2.6.29.2-tmb-server-3mdv

7. S’assurer que /etc/lilo.conf est correctement configuré pour booter par défaut sur le initrd généré et le vmlinuz correspondant
8. lancer “lilo”
9. S’assurer que le module de la carte réseau est présent dans/etc/modprobe.conf:

   alias eth0 r8169

10. s’assurer que l’interface réseau est en DHCP (drakconnect)
11. repasser la dedibox en mode normal par la console de gestion et croiser les doigts

Bonne chance!

Argh. Après avoir retenté l’expérience il s’avère que ce n’est pas si facile. Voici quelques pistes qui pourraient aider:

• Faire une mise à jour de la distribution depuis le chroot, en remplaçant les sources urpmi par une version plus récente, et en lançant une mise à jour manuelle (urpmi –auto –autoselect)
• regénérer les dépendances de module du noyau. Par exemple depmod -a 2.6.29.2-tmb-server-3mdv
• vérifier la configuration udev pour l’interface réseau qui est liée à une adresse MAC dans /etc/udev/rules.d/61-net_config.rules

¿Un pan con verdadero sabor a infancia, que se conserva por màs de 12 horas, que de buen olor a toda la casa?

¡Es posible!

Los Ingredientes

• Levadura Fresca [1] (20g)
• Harina (500g) (Puede ser cualquier tipo de harina o mezclas: blanca, negra, con semillas…)
• Agua tibia
• Sal (una cucharita rebosante)
• Aceite de Oliva (algunas cucharas, opcional)

La receta

1. Diluir la levadura con un poco de agua tibia, de harina, y una pizca de azúcar (opcional).
2. Dejar reposar la levadura en un lugar tibio hasta que haga burbujas.
3. Mezclar los ingredientes (harina(s), sal, levadura).
4. Ahora viene lo mas delicado en mi experiencia: se trata de añadir justo la cantidad de agua que haga falta, y no se puede saber de antemano porque depende mucho del tipo de harina. Lo que hago es añadir el agua tibia poco a poco mientras voy amasando la masa. Al final hay que tener mucho cuidado, esperando que el agua este bien incorporada antes de añadir más. Al final la masa debe ser bien blanda, pero no debe pegar a las manos. Si se pega añade harina.
5. Siga amasando la masa por lo menos 10 minutos, preferiblemente sin “romper” la masa. Lo que hago es “plegar” la masa de un lado y otro. Esta operación sirve para liberar un componente de la harina que va a dar consistencia al pan. Si no se deshace.
6. Separar la masa en varios trozos si quiere hacer panecillos en vez de un pan grande.
7. Haga una(s) bola(s) con la masa, coloque en un (o varios) recipiente grande y cubrir con un paño húmedo, dejar reposar… algunas horas, hasta que haya duplicado el volumen. Va mas rápido si hace calor (>25°C)
8. Amasar suavemente la masa (sin cortarla) sólo para suprimir las burbujas de aire que se formaron adentro.
9. Formar el pan, si la masa es lo suficiente firme no se necesita molde, sobre una placa para el horno
10. Dejar leudar por última vez el pan, en atmósfera húmeda y tibia. Lo que hago es ponerlo todo ya dentro del horno, con un recipiente pequeño con agua hirviendo. El pan debe crecer de manera significativa, aproximadamente 3 veces su volumen. Cuidado de no esperar demasiado, sino la levadura se muere, huele malo, y el pan cae.
11. hacer unos cortes en el pan
12. Cocinar a temperatura alta (210°C) hasta que esté bien dorado. El tiempo y la temperatura dependen del horno y del peso de cada pan.
13. Dejar enfriar antes de comer aunque huela demasiado bueno

Como se da cuenta, muchos parametros no se pueden controlar. Se trata mucho de practica. Me llevó muchos años poder lograr pan bueno, de manera consistente… ¡Pero vale la pena!

Buen Provecho.

[1] En vez de lavadura, yo uso levadura madre casera. Es mucho mejor (para mi gusto) y barata que levadura industrial. Para el modo de preparacion vea http://www.gebirg.com/Panaderia/PREPARACION%20DE%20MASA%20MADRE.htm En este caso se usa màs (como dos cucharas grandes) y no se necesita diluir, se agrega a la harina directamente.

From Word Processing to XML Writing

Editing XML for writing technical documentation has become easier recently thanks to efficient WYSIWYM (What You See is What You Mean) editors like XXE, XMLSpy, Serna, etc.

This has allowed the diffusion of XML writing to non XML savvy populations, that were glad to get away from word processor limitations and headaches. The problem came at publication time.

It is sometimes difficult to introduce the notion of “styles” in a word processor, so explaining people they have to learn XSLT and XSL-FO, plus manually edit the XSL by hand… go figure.

How to Control Presentation

Some (proprietary) tools provide helpers:

• Oxygen for example presents the list of parameters so that modifying a customization layer (DocBook for example) is easier. However all parameters are listed in no order and no check is made on parameter value.
• Arbortext provides a word processor like interface allowing to precisely control of the rendering of each XML element.

Otherwise you are doomed to learn XSL, XSL-FO, plus the documentation for each of the XML languages customizations layer you indent to use.

Towards a Standard Way of Writing Customizations Layers

I believe that, if we have a standard and structured way of writing a customization layer and its documentation, it would be easy to have tools able to:

1. parse the customization layer structure and elements
2. build a graphical interface based on this information
3. guide the user in defining the parameters
4. generate a customization layer containing the user defined data

Customization Element Types

There are probably not much element types:

1. Specific parameters, whose value is constrained by a specific type (Boolean, numeric, choice list, etc.)
2. Specific XSL-FO attributes for:

• page sequence masters
• blocks (attribute-set)
• inline

The idea is not to allow the user to define all possible rendering variable, but provide him with a structured, documented, set of parameters so he can easily make changes on the parameters defined in the “meta”-customization layer.

In Practice

2 possible implementations occur to me:

1. Meta Customization: use a meta-customization XML vocabulary for defining parameters, organizing them in sections, setting the type, the possible values, etc. So that an application can parse it, build a graphical interface, generate XSL code.
2. Add the required information for each parameter in a different namespace so the stylesheet itself can be used by the XSLT processor, and at the same time be parsed and modified by the graphical interface.

Conclusion

With such a standard, every tool that supports it would be able to show its users a nice way of customizing the transformation of all the vocabularies that would provide compatible customization layers (DocBook, DITA, TEI, …).

I am convinced it would make the life of many users easier, and many experts too

There is a tool that allows to locally mount remote DAV repositories under Linux, so it appears in the local directory tree. http://dav.sourceforge.net/

I had a really hard time trying to sort out how to make it work, given the scarce information, somethimes in contradiction with the actual system behavior. So this is a minimalistic procedure to get it running.

Note: This was on Mandriva Linux 2007. Our tutorial is aimed at mounting WebDAV URL https://dav.test.com/share/ by a simple user (“username”) so that he can actually upload files on server.

In procedures below we use # to note commands to be run as root and $ for UserName’s commands.

Preparation

1. Install davfs2 package
2. # modprobe coda

Test root mounting

1. # mkdir /mnt/DAV
2. # mount.davfs2 https://dav.test.com/share/ /mnt/DAV
3. You should now be able to browse the /mnt/DAV and alter files in there. If not verify your server configuration.
4. # umount /mnt/DAV

Important: There must be no “/” at end of URL. Note that if you use an alias on the http server for this WebDAV repository, it must not have an ending “/” on the server configuration file….
Allow users to mount the repository

1. Add username to the users group, and relog it if needed
2. $ mkdir ~/.davfs2/ ~/mnt/DAV/
3. # cp /etc/davfs2/secrets ~username/.davfs2/
4. # chown username: ~username/.davfs2/secrets
5. Add the credentials to ~username/.davfs2/secrets following comments in there:
   https://dav.test.com/share username SeCret
6. As root add the following line in /etc/fstab:
   https://dav.test.com/share /home/username/mnt/DAV/ davfs2 user,noauto 0 0

That done, UserName should be able to mount the repository using:

mount /home/username/mnt/DAV/

More automating

I was fed up by having to resize and then manually upload all images to flickr web interface by hand…

So I wrote the little script below, that calls the JFlickrUploader http://sourceforge.net/projects/jflickruploader

That script transforms images found in current dir! (use copies!):

flickrize.sh -family

by default images are reduced to 800×800.

#!/bin/bash

# http://sourceforge.net/projects/jflickruploader
JFU=/usr/local/bin/jflickruploader-1.0-rc3/JFlickrUploader
max_size=800x800
echo "Resizing images to $max_size maximum...."
files=$(find . -maxdepth 1 -type f |grep -Ei 'jpg$|jpeg$|gif$|png$' |sort -r)
for i in $files; do echo $i; mogrify -geometry $max_size "$i";done
$JFU $* -files $files

Note that first time the script is used you’ll have to register the uploading tool to flickr. Follow instructions.

Enjoy!

Modern cameras allow to take B&W pictures, but the pity is that you lose the opportunity to have the color one, unless you save as RAW as well.

Furthermore quality is not excellent in JPEG format.

However converting from color to B&W is not simple as this article explains.

Wildman found a very nice way to achieve excellent B&W pictures from raw ppm files:

convert inputColor.ppm -fx ‘(u.g*0.66+u.r*0.33+u.b*0.01)’ outputBW.tif

The endless dilema.

Though I am convinced it is possible to achieve highly complex actions through very little effort.

The problem rose again when I had to convince a newbie it was actually easy to use the photos out of my new dSLR camera. Generally the process to extract photos through a dedicated interface is long and tedious. However it is possible to achieve all that efficiently in a single click.

Look at the code below. Thanks to that I can:

1. download photos from camera and sort them by date of day in folders
2. remove possible duplicates if I did not erase camera images since last download
3. convert RAW images to a usable format

All this in one single click!

#!/bin/bash

# Change this to where to store Photos
target=/home/multimedia/Images
camera=”USB PTP Class Camera”
date=$(date –iso-8601)
mkdir -p $target/$date/tmp
cd $target/$date/tmp
# Get all photos from camera
gphoto2 –quiet –camera $camera –port usb: -P
# Do not replace photos that were already uploaded this same day
cp -u $target/$date/tmp/* $target/$date
rm -rf $target/$date/tmp
cd $target/$date
# auto-rotate using exif info
exifautotran *.JPG
# If photos were not erased from camera since last upload, remove duplicates
for i in *.{JPG,CR2}; do
for f in $(find $target -name $i ! -samefile $target/$date/$i); do
if md5sum $f | sed -e “s, .*/, ,” | md5sum –check; then
rm -f $i;
fi
done
done
# decode RAW images if not already done ?
# for i in *.CR2; do if [ ! -e $(basename $i .CR2).ppm ]; then dcraw -w $i; fi; done
# Show them!
gimv -d $target/$date

Et voilà cela devait arriver tôt ou tard, me voici charmé par le chant du blog. Flock a joué un grand rôle dois-je dire, pour ma décharge?