This is my (not very) regular series about what I have read on the web since last time.

Jürgen has written a post asking whether in the age of mobile phones, the need for a wrist watch is diminished?

Are smartphones a complete waste of time? Bug looks into the pros and cons. K thinks the iPhone is a big con, I have to agree. However, Garrick loves his iPhone.

Justin has a cat fight over OS X 10.5 (Leopard) playing up. For my sins, I have had to use OS X a bit in my new job, and I actually found Leopard less annoying than Tiger, mainly because in each version, OS X becomes less like NextSTEP and more like Linux.

Brock tries out XMLStarlet, the command line toolset for XML processing. Daniel looks at Logical Volume Manager (LVM) on Ubuntu and Gentoo. Paul has started to set up a backup server.

Andrew W dug up a nice graphical guide to the system crontab file. I personally am very happy at whoever invented the /etc/cron.hourly and /etc/cron.daily folders which are good enough for me most of the time.

Mez reminds us of the virtues of compressed air. Danux has started a new site called Amarus, there is not much there at the moment, but we wish him well.

Andy L talks about an issue I have been thinking about before, namely, if he current world wide web gets taken over by narrow minded corporate interests, shall we start our own World Wide Web? I have a slightly different suggestion, lets re-invade the forerunner to WWW, gopher.

Recently, at a conference that shall remain nameless, some cynical but funny person made a joke about the great BDFL. He did an impression of a Guido Van Rossum doll with a pull-string in his back, when the string was pulled, the Guido doll would talk half a dozen phrases about Python 3000 (and nothing else). Interestly, Craig Balding managed to interview Guido on a different subject, Google App Engine Security, and true to the joke, Guido says almost nothing.

Django NewFormsAdmin

If you do use Django, then you will want to know that the Django NewFormsAdmin branch has been committed to SVN. Therefore, if you are running Django from the SVN version, then don't SVN up until you have changed your code.

Basically Admin functions are now not part of the models.py file but instead are in a separate new file called admin.py. So cut and paste your admin classes from models.py to admin.py as explained in this guide. This is the last major API change before Django becomes 1.0 in September.

This will presumably keep Christian Joergensen happy, as he recently had a moan about Django's release schedule, i.e. Django has not made packaged releases that often. I personally disagree with Joergensen. For this type of software, releases are somewhat arbitrary and over-rated marketing tools.

For open source software, the mainline trunk should always be in a releasable state. With distributed development (i.e. when branching is cheap and easy) then there is no need for an old fashioned cycle of plan-develop-freeze-test-release-plan-develop-freeze... The trunk should be constantly tested.

The author admits that web frameworks move faster than some other types of software:

"This is a very long time, when you're in the market of web frameworks."

So Django is not a GUI WYSIWYG web site creating program. You can't just casually pick it up and make a website, you have to put time into it. To get the most out of Django, you have to read a huge pile of (mostly well written) documentation. Even for a seasoned Python programmer who knows other MVC frameworks, it will take an evening or so.

After this initial investment, if you decide to make your web applications using Django, then you are already committing yourself to keep up with the developments and improvements in the framework, i.e. keeping up to date with what the Django developers are doing. Therefore, tracking SVN is not unreasonable if you already know what changes are coming. Almost everyone paying even scant attention to Django, would have known about the impending NewFormsAdmin, the documentation page about it that I linked to above was first published on the 14th January 2007.

I do accept however, that Django does seem more suited for teams maintaining the same websites over time, e.g. in-house programmers or contractors on long-term service agreements; rather than one-off, develop and leave type development. However, the former probably does produce better web sites.