You can download the sources from the SourceForge project.

Unofficial Debian packages are being built right now, but will only be uploaded to my repository after a bit more testing, during the week-end I guess.
They are meant to be uploaded to the official Debian archive once they are considered as ready by the pkg-dspam team.

Hopefully, they will be part of Squeeze.

161 posts, only 190 comments, but the statistics shows a stable increase in the number of visitors (in average a bit less than 500 unique visitors per day).

During the week-end, I have worked on upgrading Wordpress to the newer 2.9 release, and have just finished optimising my apache and mysql configurations. Access to my websites should now be somehow faster.

The migration has caused a downtime of more or less 4 hours yesterday, time for the data to cross the ocean.
The migration was successful, though you might still experience issues to connect to the services hosts on this linode due to the new IP adressess (both IPv4 and IPv6). Most DNS servers should however now be up-to-date.

Thanks again to Linode for their great service and support.

Ubuntu One: I wish we could have something similar for Debian users (ideally, it should be extended so that we could sync our calendars, contacts, liferea data etc.).

usb-creator: seems the easiest way to create a bootable usb stick containing a complete and customisable environment. I am aware of the existence Debian Live, but I must say I haven’t given it a try yet.

Desktop orientation: Ubuntu users are in a large majority desktop users. Ubuntu is imho desktop-oriented, and I particularly like the artwork coherence (from boot loader to default desktop).

rkhunter was moved to Alioth’s collab-maint project, and I have set up a personal GIT repository for the other packages.

This page on the wiki has helped a lot.

I now need to learn how to use this tool, but my first tests are very encouraging.
I haven’t had to change my packaging workflow, switching from svn-buildpackage to git-buildpackage.

I am even able to build i386 packages on my amd64 machine as before without the need to change anything in my ~/.pbuilderrc.

I first had to switch from dpatch or cdbs’ simple-patchsys to quilt, which was easy thanks to Romain Francoise’s old blog post.

Also note this quick adaptation for cdbs:

for i in $(ls debian/patches/*.patch); do \
    I=$(basename $i | sed 's/.patch$//'); \
    quilt import -P $I.diff $i; \
done

The rest was much easier than what I thought, thanks to the related page on the wiki.

I have already tried uploading a package to my personal repository, and reprepro just worked as expected.

I have had to found the right way starting it for my session, anf finally found the following was the easiest and cleanest solution:

echo export WINDOW_MANAGER=/usr/bin/compiz >> ~/.gnomerc

Writing it here so that I can find it again…

This is not a real surprise, as it has also happened with the HADOPI I law last May, but this time, I will take the time to write him a letter to express my feelings.

I will only quote the following parts of his letter, referring to the amendments proposed to this law, following to the partial censure of the Constitutional Council:

[...] ce qui me laisse plus que très dubitatif sur l’efficacité du dispositif. [...] cette procédure me semble tout à fait inadaptée.

Which can be translated as follows:

[...] I doubt this disposal will have any effect. [...] I think this procedure is absolutely not adapted.

Now, let’s bet what will be his vote in September!

For my personal email, I use a setup based on Postfix, RBL and WL, grey listing, and DSPAM (by the way, you can find unofficial packages of current development version in my personal repository).

To better train DSPAM, I have removed all other filters (except some very basic controls at SMTP time), and I am very happy of this setup (grey listing delays receipt of email, which I am not very satisfied with, and relying on third-parties RBL is also not the best thing imho). This lead however to much more mail in my DSPAM quarantine queue.
I will hence re-enable more advanced filters at SMTP time once DSPAM bayes are large enough.

I have also played a bit with SPF, and configured my DNS server for my main domain to use it. It now passes the recommended tests. However, I haven’t implemented SPF checks in my Postfix setup.

For work, we rely on a (very old) MS Exchange setup, which is about to be replaced soon by a newer MS based solution. The costs for this replacement is quite high for a small company (about €7,000 including the hardware, the required software licenses, and the costs for setting this up by an independent worker during our summer closure).

I have hence tested Google Apps Standard Edition with one of my domains. I am always very much impressed by Gmail, which is really fast, seems reliable and easy to use. I haven’t tested it carefully in “work conditions” though (ie. using MS Outlook client for calendar sharing etc.), and the cost of the premier edition is €40 per user and per year, which means €1,600 per year for 40 users, ie. it would take more or less 4.5 years to reach the cost of our replacement server. Considering the current server is about 10 years old, I am wondering if Google would really be an alternative (other factors would also play a role, like our low quality DSL connection, 10 years for a mail server seems to be much too old, considering we cannot run any good anti-spam software on it etc.).

Note that an opensource solution was not considered, nobody could administer it (this is not my function in this company).

SaaS (Software as a Service) seem to be the future of software (this is at least what major companies in this sector tend to say, like Microsoft, Adobe, Google to only list some examples), is it already mature enough for small companies largely depending on their email like us?