As a Firefox user I have the Greasemonkey add-on at my disposal. Greasemonkey can selectively modify web pages on-the-fly to do almost anything to the content like:

• Remove or hide parts of the page
• Add bits to the pages
• Insert code to periodically reload a page that normally wouldn’t.
• Convert URLs embedded as plain text to real hyperlinks (another good one for the forum)
• It can even incorporate parts of other pages or sites

What I ended up with is a Greasemonkey script that inserts a site-targeted Google search box into the forum’s tool bar. The result is shown below:

The inserted search box should fit comfortably on any screen of 1024 or more horizontal pixels. Search results are targeted to the ABC science forums and end up in the forum’s main frame to avoid taking the user away from the forum. As a bonus the search box also appears on the TechTalk tool bar.

If you are interested in trying it out then:

1. Ensure the latest version of Greasemonkey is installed.
2. Click this link to install my script.
3. Tell me how much you like it

As an IT literate guy I frequently find myself explaining to others how even the simplest IT things can be done more efficiently (or effectively) with small changes in behaviour or thinking. Often these things are as straight forward as, “You can copy and paste using Control-C and V” or “There are better tools than Excel for collecting this information.” Some people take these things and run with them, and others revert to type within minutes but this rarely leads to the sky falling in. This level of individual coaching, however, is not going to attract work at $100/hr.

Sometimes more complex offerings involve the other party rethinking how they approach what they are doing in order to facilitate the types of efficiency gains that Seth refers to. This poses quite an obstacle in large project/organisations. A common scenario is a project sub-team that is focussed on how the output of a process looks, e.g. a report in given format, and considers that the only way to do it is to collate information, in isolation from others, into that output format. This, IMHO, is rarely the most efficient way to do it even in the unlikely event that their outputs are independent of all others in related tasks. However, addressing any IT efficiencies in this process are almost always focussed, by the client, on ways of collating information in that output format and not on a more holistic approach to the problem(s). My experience of several large Defence and commercial projects is that the larger the project the greater this inertia of thinking becomes. Of course, large projects are often supported by IT teams that should be guiding the project’s computing directions: in practice I think they are often marginalized by the project management (outside of IT projects they’re running) or organisational issues. Complex environments also dictate a longer required (non-productive) period by the coach to assess the most effective changes; creating a financial barrier that feeds the thought inertia. Whether an external digital coach would have any effect in this environment is debatable.

I am sure that Seth’s notion of ‘digital coaches’ will work in some circles. It has a greatly improved chance of success in smaller settings where the coach is requested, respected, and has the ear of someone with the desire and capacity to enforce/ensure adoption of change. It seems to me essential that the coach can grasp the situation in a reasonably small time and start making productive suggestions: after all the customer does not want to be paying $100/hr for two weeks to see some result. Seth’s example of a restaurant or small business seems to be about right. Dave Saunders commented, “I love the idea of a digital coach. I sure wouldn’t want to do it though.” I agree: at least until I’ve tried it in a small business. I’ve already tried it in large organisations, and the best I could say is that it’s often demoralising and usually ineffective. Maybe that’s an indictment of my own abilities rather than the concept.

One of the large figures in the free software world is Eben Moglen, a professor of law and legal history at Columbia University and heavily involved in defending and extending the GNU General Public License. In the video below Eben Moglen is speaking at a lecture for the Scottish Society for Computers and Law titled, “The Global Software Industry in Transformation: After GPLv3.” The video is 80 minutes long (about 160MB) but Moglen manages to eloquently encapsulate why Free Software is important in the first 20 minutes. If you are in the slightest interested it is well worth the time.

The event is also available as streaming audio and a transcription.

I found this talk one of the most persuasive arguments for the freedom of software (and knowledge in general). In addressing the reproduction of knowledge in digital form Moglen notes that the incremental cost of further copies is approaching zero and follows:

The consequence of those changes is the onset of a very powerful moral question. If it is possible, easily possible, to give to each human being who wishes it, anything of utility or beauty in our world of civilisation, if it is possible to deliver any such entity anywhere at any time at low cost or at zero cost, why is it ever moral to exclude anyone from anything she wants?

This argument really appeals to me.

References:

1. The video is sourced from The Internet Archive and is released under the Creative Commons: Attribution-Share Alike 3.0 by Al Broom, Valley Technology.
2. The GNU image is by Aurelio A. Heckert, release under the Free Art License. From http://www.gnu.org/graphics/heckert_gnu.html

Many (most) motherboards have an ability to wake up at a certain time, but this must be manually set from the system setup pages. I initially tried the NVRAM Wakeup utility. This utility works by directly writing wakeup time and date information into non-volatile memory and is very tightly coupled to hardware and software revisions. My motherboard (an A7V880 BIOS rev 1008) was supported out of the box and it seemed to work. Unfortunately though, this motherboard has other issues that I needed to remedy with a BIOS update (to 1009.001). This meant that I had to go through a repetitive routine to have NVRAM Wakeup guess the memory locations of interest. I did get a working setup but I had to manually hack the automated guess, and even then there were elements of doubt. Every so often the machine would wake to corrupted NVRAM contents . While I am not certain that the wakeup routines were the cause I thought I’d migrate to a method that was marginally less risky.

The Linux kernel ACPI implementation exposes the BIOS alarm system as a file /proc/acpi/alarm. If you look at the content of this file is gives a human-readable representation of the currently set alarm time like 00-00-05 21:30:00. By writing to the file in the same format you can set the alarm time although usually the month and year are ignored, and 00 in the day field means any-day. The time must be written to the alarm file in whatever time zone the hardware clock is set (UTC on most Linux-only machines, but sometimes local time if the machine is shared with Windows). There is another way to write the time that avoids needing to know whether the hardware clock is UTC or not. Writing time in this format +00-00-01 02:45:00 will cause the kernel to add the specified days, hours, minutes, and seconds to whatever the current time is and store that in the alarm. I wrote the script linked below to do just that. It acts as a drop-in replacement for nvram-wakeup and accepts the same options and inputs (although it ignores most of them). The desired wakeup time is passed from MythTV as a UNIX time stamp: seconds since 1970-01-01 00:00:00 UTC. Appropriate mathematics is done and a time delta is stored into the alarm file. The script is limited to times within 31 days, which should be adequate for most purposes. This script has been working well for the past month.

I believe there are two other gotchas that might be of interest:

• On some motherboards the alarm time is reset if the hardware clock time is updated during shutdown. This is usually in a script halt.sh (or the like). You can work around it with something like:
  BACKUP_TIME=$(cat /proc/acpi/alarm)
... the existing clock setting code (hwclock)
echo $BACKUP_TIME > /proc/acpi/alarm
  to restore the value after the clock is set.
• From kernel version 2.6.21 or 22 this support is likely to disappear and be replaced with less PC-centric clock handling. When this occurs I will be rewriting the script.

Now, if only I could perfect the commercial skipping feature

Clear skies!

Download ACPI Wakeup Script

The MythTV screen shots are from the very ornamental Blootube wide theme by Justin Hornsby.

Written in glorious Fortran 77, SWIFT could be built with many commercial Fortran compilers including Intel’s Fortran Compiler. I successfully built SWIFT using the Intel compiler although I never used it in anger. I was never able to build it with the GCC 3.x Fortran (g77) because of the code’s use of recursive routines. Two things have changed in this regard:

• GCC 4.1+ now includes gfortran, which aims to be Fortran 95 compliant.
• David E. Kaufmann has authored a completely redesigned version of the Swift package with the support of NASA’s Applied Information Systems Research Program (AISRP).

The enhanced package is called Swifter and is written to Fortran 90 standards. It now compiles with the GCC gfortran compiler with the following change made to the Makefile.Defines:
FORTRAN = gfortran
FFLAGS = -O -ffree-line-length-none

Now I can attack the problem of building funky animations.

Clear skies!

After recent upgrades to my home PC I discovered, to my great dismay, that Kobo now ran with terrible sound glitches and crackling on Windows, and not at all under Linux. The new sound system is a Realtek ALC883 (Intel high definition audio) and most everything else worked properly. Fiddling with the range of sound options in the game did nothing to alleviate the problem.

The game uses the SDL for its sound, and ships with SDL libraries. I started looking at upgrading the SDL library and found the shipped version were a few dot-releases old. The following steps update the SDL libraries in Kobo Deluxe on Windows:

1. Download the latest Windows SDL library binaries
2. Download the latest Windows SDL-Image library binaries
3. Unpack both archives
4. Copy all the unpacked DLL files into the Kobo Deluxe game directory, overwriting existing files.
5. Play!

This seemed to fix the problem on Windows.

On the Linux side of things, the game hangs trying to start with audio enabled. This problem may also be related to using a dual core processor. I’ll keep you posted with any progress.