Jangle Jack

I’m starting in on a snazzy workbench project which will encode a bunch of it’s working data in the METS schema. In the process I’ll probably need to understand and use the standard in ways that have been only theory to me. Premis in METS, also exciting for me.

In any case, I found myself constantly flipping back to the Oxygen XSD viewer in Eclipse to consult the standard.  That’s very cumbersome to do all day long. This one time I tried opening an additional workbench window within the same Eclipse instance, using Eclipse on two displays. Nice in theory, but I do not recommend doing this if you want to retain a stable Eclipse IDE.

That’s when I noticed the plotter sitting in the GIS area here. It can print 3 feet by whatever length! The next thing I discovered is that you can save the current Oxygen XSD schema view to an image, capturing whatever parts of the schema are expanded. (Thank you Oxygen.)

The first draft I tried was an image with the basic parts of the entire schema expanded underneath the root element. Even after collapsing all repeated element types, this was a ten foot long poster. While fun for demo purposes, this would be impractical without inventing an analog scrolling device. I was able to shorten the poster to a more reasonable length by breaking the schema up into element types and rearranging to use the 3 foot width. Now the poster is 3 feet wide by 6 feet tall, small enough to pin to the back of a tall bookcase or cube divider.

Show them how serious you are about metadata transmission!

If you use METS, I highly recommend you look around your institution for a plotter and go to town. The final image and Gimp file are linked.

I’ve been designing Fedora content models for two years now.  The whole time I’ve struggled to communicate the mix-in approach to content models.  This is a variant on the “decorator” design pattern that’s familiar to object-oriented programmers, but pretty unfamiliar to most others.  The models I used for the repository at UNC do not have inheritance, just a set of models that might or might not apply to a given object.  So at least the diagram doesn’t need to express relationships between models.

In any case, the other day I saw that folks at Wisconsin had an elegant solution, color.  Follow the link and look at the PDF diagram.  It seems like a small thing, but for expressing content models it is a real breakthrough.  You can lay out your example objects in a normal-looking composition diagram, then color code the content models to match the traits that they imply.  I don’t know how decorators are modeled in UML and thankfully now I don’t need to…

Now the Wisconsin model is beautiful, with drop-shadows and curvy shapes.  I want that too, but I prefer open source tools.  So that limits me to only a few diagramming apps, namely Dia, OpenOffice Draw, Inkscape.  Okay not so few.  My preference is Dia, of all my options it is closest to Visio and I’ve done a lot of work with it in the past.  However, there weren’t any shapes quite suitable to the task at hand in Dia.  (There probably aren’t in Visio either..)  So now I had to roll my own Fedora modeling shape.  Like most such things, it really doesn’t take very long once you find the right documentation:

• http://projects.gnome.org/dia/faq.html#AddingShapes
• http://projects.gnome.org/dia/custom-shapes

I found the best way to create a shape was to make a template from some similar shape that ships with Dia.  Find your Dia preferences folder.  On Linux it’s in a .dia folder under your home directory.  The .dia folder (or equivalent) contains a shapes folder and a sheets folder.  So I created a new subfolder in each of these locations called “Fedora”.  Then I copied a sheet file and a shape file into these locations from similar locations in the main Dia install.  On Ubuntu linux Dia installs it’s default shapes and sheets in /usr/share/dia.

I created the outline of my new shapes in Dia and then exported them as SVG.  This SVG went into the shape file.  Then I customized it based on the second link above.  I ended up with two basic fedora shapes, one for objects and one for datastreams.  They are a work in progress, but allowed me to create the diagram below.

click for full size

There are some lingering todos, mostly making the arrows able to link to the center instead of an edge.  That probably involves replacing separate SVG lines with a closed path description that can be filled.  I also want to find a way to depict model-tied services, something like UML class operations I suppose.  Or, maybe someone else will pick up where I left off.

Here’s the shape/sheet bundle:   fedorashapes

Maybe I’m being a nasty post-colonial oppressor and shoving aside Caribbean sovereigns, but I have a problem with calling this fella a knight.  So this Texan is knighted in Antigua/Barbuda, so what?  He basically owns the place.

HE IS THE LARGEST EMPLOYER IN ANTIGUA AFTER GOVERNMENT

If found guilty of these fraud allegations, then they should dub him privateer and send him forth to scuttle rival ponzi schemes in Grand Cayman.  Ah yes, echos of Python wit on corporate raiders.  Let’s just call him “Great Pirate Allen” and then the ladies can coo with delight every time they say it.

If you work in higher education, Blackboard is the 800 pound gorilla. Now open source advocates are challenging 44 of their patents. In an effort to gather evidence of previous work they created a fascinating Wikipedia article on the subject. See History of Virtual Learning Environments. I think Douglas Engelbart is one of the more forward thinking people in this history. Invented hypertext, the mouse and early user interfaces. There is a video I’ve seen of him running a hypertext demo back in the late sixties. If you watch it, wait for the title cards to pass and you will be rewarded.
Anyway, I found the Blackboard news refreshing.

This post from The Content Wrangler looks at the future role of documents in the workplace and who will be called in to analyze and improve the situation. It reviews a book that might be a good add for our KM wishlist.