So I continue my regular series of what I have read this week. Not a lot in fact as there is near silence in my RSS reader, it seems that everyone is on holiday and people are not blogging too much. However, a few people had some interesting things to say.

Organisations for Britain's first ever Python conference are continuing and seem to be going well, see you in September if you are going.

Thanks to Gentoo Linux Newsletter who kindly featured my blog in the latest issue.

Thanks also to Slashdot who linked to my blog this last week. Actually the server did quite well considering there were over ten thousand unique visitors in half an hour, it was a little bit slow to start with but never actually stopped giving out pages and remained fully responsive apart from Apache. Below is an image of the top command at the peak (click for larger):

`.. image:: http://commandline.org.uk/images/posts/other/top-s.png

System Message: WARNING/2 (<string>, line 19); backlink

Inline interpreted text or phrase reference start-string without end-string.

alt:	Processor 97.8%, hardly any free memory

`_

Anyway, by now, you probably all know that a US court ruled that Novell, not SCO, owns the UNIX Copyrights. I was going to link back to all the places online over the last four years where I had said that SCO had no chance, but that would be too easy. I made a joke in 2003 about short selling SCO stock, but obviously didn't since I had no money and no stockbroker. Perhaps I should have done, I would have had 20x returns. Well done to anyone who did!

Phill has been using Python with MySQL, beware it is MySQL as you know it but there is one more task you must commit to. Brian wonders whether it is less fun to write a program in Java or have your eyebrows burned off.

Bug reports that Microsoft censor links in his instant messages, I wonder what else they are doing with your conversations?

Dan Ballard talks about Hardened Gentoo, I have used it on my home router/file store and found it quite easy to get an extremely over secured system together.

Scott ran into the problem of bouncebacks from spam messages that pretend to be from your domain. I got this also a few years ago, and like Scott, I just enabled only the mailnames that I need. I suppose there is not much else you can really do about it.

I am trying to give up Mac hardware, they are a bad addiction, but looking at Garrick's visual comparison of a iMac and a Dell almost pulled me back into the Borg.

Ciarán is organising a software freedom stall in Birmingham city Centre, and he also takes nice pictures of butterflies. Ravi finds a free software sticker book.

How many Linux users are there?

According to the visitor statistics of this website, Linux use is over 60%. This is obviously not representative and I have yet to be shown a sample that can be well defended methodologically. Website statistics are a very dubious way of counting usage as different websites will appeal to very different demographics, and Linux take up is a complicated phenomenon.

Roy Schestowitz has a good think about these issues in "Can Linux Adoption Ever be Accurately Gauged?". I agree with Schestowitz that all current attempts to guess the current number of Linux users are not credible if we look at them with an ounce of statistical knowledge.

However, my view is that the people best situated to compile this information (the Linux distros) have been woefully ineffective at it.

It is easy for the distributions who sell support to count licences of course, the accountants will figure out that Redhat have sold x millions licences per year and Novell have sold y million licences per year. However, they will have users who do not want paid support, and these are probably the majority of their users.

What I would say is that only live boxes ask for updates; so I personally would track every copy of each version of a few select packages. The package manager itself will be in all boxes, so I would track the package that updates apt-get/portage and so on.

The next step is to try to distinguish between desktops and servers, so I would track the total numbers of gtk+, QT, Apache and Samba. You might tweak the list slightly, but this kind of information, collected and presented in the aggregate, is hardly intrusive.

Of course each distro has to collaborate with the maintainers of dozens of mirrors, so it would be a major undertaking, but we need to start thinking about having an accurate and credible count to help campaign for hardware support and so on.

Novell did an interesting survey of 27,000 OpenSUSE users, which shows a bit of who they are (guys in their twenties) and where they are using OpenSUSE (mostly at home). When I get a spare moment, I will read the full report closely and see if I can find others too. There should be enough information out there to gain a reasonable understanding of demographics and how people come to Linux, after that we might be able to think more about the methodology of counting the installed desktop Linux base.