geekroom

I managed to mess up the upgrade of our webgui system yesterday morning, so I’ve spent much of today fixing the mess (mainly just missing form labels in a bunch of admin screens). During the updgrade, I missed one of the database update scripts, so I’ve been manually running the queries that were missed all day, doing my best not to break anything in the process. I managed to nuke half of the site templates at some point (including the all important article templates!), but I quickly restored them from a database backup.

Everything seems to be working well now, and I fixed a couple of webgui bugs in the process, so hopefully the site is better off for the effort.

I had to reboot the blogs server to install a security patch tonight, and of course nothing went smoothly.  The stupid IBM server refused to boot without a keyboard, and we don’t have a keyboard attached (why the heck make a server default to not booting without a keyboard?!?!).  After driving in to the server room and telling it that I really did want it to reboot, I still had to reboot the machine twice to install all of the windows updates.  Ten years later, why are we still having to do the reboot dance on windows machines?  Once is bad enough for a production server, but twice, with one of them taking about 10 minutes to do the actual update on a reasonably quick machine?  Suboptimal.

The world seems to be happy now, no data was lost, Harvard is blogging again, and I’m the proud owner of two more white hairs.

I’ve setup a webgui instance at http://cyber.law.harvard.edu/apps for random small applications, presumably within other sites. Please use this system for any small webgui application that does not require its own userbase (a unique username requires a separate database and therefore a separate webgui instance). For example, I’m implementing the BloggerCon registration system on this site. When creating a new application, please place it within its own, new root. See me to get an admin user setup.

I am officially a huge fan of the Template Toolkit module for perl. I’ve been developing an online survey application recently using the TT stuff along with the CGI::Application module, and it has made writing a medium sized perl web application astoundingly easy. I’ve been able to spend hugely more of my time writing actual application logic rather than stupid infrastructure code, and the infrastructure has held up easily as the application has grown from a few hundred lines of code to several thousand. It manages this trick without all of the overhead entailed by using something like Java Struts, which is great for large applications but overkill for medium sized stuff. And the templates written with the TT language are vastly more enjoyable to deal with than jsp (which I like less the more I use it).

This is a boring documentation post!

If you want to clip part of a Real Media file from a larger file (or make changes to the Title, etc. meta-information) you have to use a command-line utility which is provided with RealProducer. This is called rmeditor.exe, and it is located in the same directory as Real Producer itself. Here’s an example of taking a minute-long chunk out of the middle of a .rm file (on a CD-ROM in this example) and making it into a new clip:

1) Open a command line (Start menu -> Run… -> cmd)
2) “C:\Program Files\Real\RealProducerPlus\rmeditor.exe” -i d:\input_filename.rm -o output_filename.rm -s 0:0:3:2.20 -e 0:0:4:2.20

The numbers after -s (“start”) and -e (“end”) are times, expressed as day:hour:minute:second.millisecond. And that’s it! A minute-long file named output_filename.rm will appear in the directory you ran the command in. This command can be used to join different .rm files together, and other tricks as well – see Real’s documentation for more details.

The adaptor I got to mount a 44-pin laptop IDE hard drive on a desktop machine works perfectly. It took a bit of finagling, though, since it didn’t come with any documentation, nor does the manufacturer provide any I could find online.

So for future reference, here’s a picture of both sides of a similar device:

Note that on each side of the adaptor, pin 1 is on the opposite side from where the power line goes in. This is counterintuitive if you’re guessing based on IDE drives, where pin 1 is usually next to the power.

The power line plugs in where the adaptor case has “R” and “B” on opposite sides – R is Red, positive, and B is Black, ground.

Of course, to mount any laptop drive on a desktop system, you still will have to figure out where that drive’s pin 1 is, how to set master/slave, etc.

We own a Sony “NetMD” mini-disc walkman, model MZ-N707. I tried using its handy “microphone jack” to record Phil Malone’s lunch lecture yesterday (thank you Whitney for the nice mic!). The recording is fine, I guess – or at least it sounds good on headphones.

But despite the logo promising MP3 compativility [sic] and the USB port, there is no way to get data off the recorder in a digital format. The only way to talk to this walkman is through the Sony-provided “OpenMG Jukebox” – a DRM-encumbered monster which only allows you to “check out” songs from your hard drive to the walkman. Once they’ve been “checked out” to the walkman, there’s nothing you can do with them except “check them back in” to the “OpenMG”. And you can’t “check in” anything you didn’t “check out”. So the digitally recorded audio sits there on the walkman and the only way to get it out is through the headphone jack.

That’s wicked useless!

We had a good laugh up here about the irony of the name “OpenMG” for such very closed software. And about “NetMD” – you’d think that might have something to do with sharing out information over some sort of “network” but you’d be wrong. But many others have made the same mistake, as googling “mz-n707 broken” or “mz-n707 drm” will quickly tell you.

The moral: don’t use this stupid dumb device! Many of the bitter things people have written about this deliberately broken device suggest useful products you can buy instead, if you’re in the market.

1. start up the exhange administrator program
2. go to tools->directory export
3. select a file
4. uncheck recipients and check distribution lists
5. export
6. grep ',team,' team.csv | awk -v 'RS=%' -v 'FS=cn=' '{print $2}' where team.csv is the exported file and team is the desired list

This just gets the recipient nicknames. To get the actual email addresses, you’d have to export the recipients as well and write a more sophisticated script to lookup the email addresses from the nicknames.

We’re currently waiting to get the licensing issues figured out before releasing the 1.1 version of H2O. In the meantime, we’re pondering what to put into the 1.2 release. Here are the current leading candidates:

• threading for the message boards
• cross project message board discussions
• project announcements
• optimization of the interesting projects query for the front page
• project sections
• section only rotisserie and message board discussions
• batch user import (csv)
• misc small bugs

As these things tend to be, this has turned out to be more ambitious than hoped. Most of the features are required for an experiment we’re hoping to perform with our fall HLS courses that will test the quality of discussion produced by rotisserie vs. standard message board discussions.

The excellent Maggie Cohen has been making incredible progress cataloguing the Center’s huge archive of multimedia materials at the Berkman Center Media Library. She’s worked through nearly the entire library, describing every lecture, conference presentation, picture, and file present in the library and adding person, project, and other categories to the pieces. Once she’s done, we hope to CC most of the material and promote it more actively as a living archive of all of the great material produced by the Berkman Center over the years. If you haven’t looked at the media library recently, take a look now. Especially try playing with the text search and the category based navigation, which both now return meaningful results.