I haven't blogged for some time, and checked out my web statistics to see if there's any real people visitors left, or just robots. Surprise, surprise. Lots of folks keep finding their way here. Welcome!
But the biggest surprise was to the end of the list of operating systems used. 14 Visits in november using Windows 3.1. Windows 3.1! That's like: man! That is ages ago!
That even predates the invention of the information superhighway by Al Gore!
I'm old enough to remember those days... Trumpet Winsock on a 14 kilobit dial-up.
So, I also check the browser types used. Sure, there they are: Netscape 4.7, Internet Explorer 3.x, Internet Explorer 4.x. Dinosaurs roaming the social net in the 21st century.
Reflect on this for a moment. Windows 3.1 was superseded by Windows 95 in, exactly, 1995. Fourteen long years ago. I don't know what is more amazing: that there's people who are actually interfacing with such prehistoric software? Or that there's machines that still haven't broken down after more than 14 years?
Or that these systems haven't been totally barfed by malware by now? Maybe these systems are so old, that they've become, effectively, immune from malware? In the sense that the malware genome just doesn't fit their vulnerabilities anymore?
Anyway, there's some important lessons there. First, the long tail. Even if you run a relatively obscure website, you'll get visited by any browser ever made, running on any computing platform that has ever been internet-enabled. Even if you thought those technologies were long extinct already.
Second, path dependence. The headlong rush of technology progress seems to leave history far behind, as long as you thunder along in the vanguard. But if you take a distance and look carefully (like in the long tails of your server logs), you'll see that each of the steps made along the way keeps having impact on the present. In a creepy way, the past refuses to die and keeps reaching out for us in unexpected ways.
So, dear reader, take care. All those decisions you make, assuming some sensible implicit "use by" date? Some automatic shelf life that in all reasonable circumstances should limit the scope of any (unforeseen) consequences? Wrong assumption. The past is forever with us, and the present will be too. We are not only creating the future. We are also, right now, creating the past of the future. The future past that will remain forever present.
This must be the stupidest ad I ever saw on Slashdot. The biggest anti-patent nest I've ever seen and Google's hawking a patent broker....
Though, come to think of it, you will see a lot of clickthroughs I guess, from slashdotters eager to do a slashdotting of the advertiser's clickthrough budget 
. So, from Google's AI point of view, that's a successful advert: the clicks keep on coming.
One of the interesting things about the upcoming Plone 3.3 release is
it's
consistent implementation of INavigationRoot
In short, this means it's easy to create a design where the portal tabs (the "global sections") do not need to be toplevel folders in your site. I need this for a design I'm currently implementing, where the horizontal "main" navigation is actually the second-level navigation.
Now, instead of rewriting the complete global_sections viewlet
PLUS the webcouturier.dropdownmenu integration... all I need to
do is mark an intended "toplevel section" as such by applying
the INavigationRoot marker interface.
There's a content type
NavigationRootFolder
that does just that. But it's not properly eggified.
It's easy to write your own custom contenttype that subclasses
ATContentTypes.content.folder.ATFolder and apply the INavigationRoot
marker interface to it. But still, you need to do all the plumbing.
Luckily, we've got p4a.subtyper. That's a nifty little component that's
been designed precisely to apply marker interfaces on content objects,
run-time. This allows us to be lazy, which is A Good Thing®.
What we're going to do, is set up a marker interface that allows us to turn vanilla "Folders" into "Sections" that behave as-if they're subsites.
Indentation in emacs css mode was driving me crazy. Luckily I found this blog entry who quotes this blog entry who quotes yet another source that is offline now.
These settings didn't quite work though. CSS mode had to be loaded, before setting the cssm variables worked.
The resulting block in my .emacs:
;; fix css mode (require 'css-mode) (setq cssm-indent-level 2) (setq cssm-newline-before-closing-bracket t) (setq cssm-indent-function #'cssm-c-style-indenter) (setq cssm-mirror-mode t)
This still gives an error on the first edit on a new css file. But works as intended in normal circumstances. Back to work now
Last week, the nice folks at osor.eu invited me over to do a usability review of their website.
OSOR (Open Source Observatory and Repository) is a multi-year project by the European Commission to promote open source software use in public administrations.
We were able to have a very frank and constructive dialog about usability issues with the current website. The site is a Plone CMS frontend coupled with a Gforge software repository.
There's a couple of design inconsistencies in there that are deeply rooted in the project's mission concept, and fixing them runs into policy constraints. Meaning, it can't be fixed right now.
This confirms my experience, about the way that organizational fuzziness tends to show up as design inconsistencies. Which then translate into technology and usability problems further down the line.
Creating a truly open dialog that connects technology design with business strategy, results not only in better designs, but also improves organizational learning.
To date 40 years ago, Apollo 11 was on the way to making history. In his fascinating study Digital Apollo - Human and Machine in Spaceflight, David Mandell traces the development of computing technology supporting the Apollo project.
The book opens a window on the birth of software engineering as a discipline. Some stuff is highly unfamiliar, like programming with "core ropes" .
The permanent memory, which stored the flight programs, consisted of a complex series of wires running in and out of magnetic cores that determined if a particular bit in a memory location was a one or a zero.
Other issues are shockingly recognizable after all those years.