This is a follow-up to yesterday’s blog posting that described the experience of upgrading a Drupal site from version 5 to version 6. To see the raw notes from yesterady, read the original post here. After all was said and done yesterday, I had (mostly) successfully upgraded Drupal 5 to Drupal 6 using our development servers. This involved updating or abandoning about two dozen modules, and took about 3-4 hours in total. However, we had two errors on the site:

1. User headshots were showing broken links
2. The status report page was showing a message that “HTTP request status fails”

A quick search resulted in no promising resources for the first problem, but the second problem turned up this tread in the Drupal.org forums. Reading through the several dozen posts, there were a handful of potential solutions posted, depending on whether the Drupal site is using a proxy server (we aren’t), running on a localhost test box (we aren’t), etc. The one solution that came up most often, even across some causes, was to edit the etc/hosts file.

Specifically, the recommendation was to add the IP address and name to the etc/hosts file, to help the site query itself – otherwise there is a apparently a risk that the self-query will resolve to localhost rather than the site name (www.foo.tld), which may throw an error.

Another cause of this problem is apparently having the site in maintenance mode – which is part of the upgrade process.

The bonus: once I edited our etc/hosts file to add the IP and name of this site, not only did it solve the “HTTP request status fails” message, but the headshots also started working! Woohoo!

(For the record, there is also a HTTP Request Fail Reset module, but the author’s notes are that this should never be used on a production site – so I didn’t want to rely on that for a solution)

Today (August 12th, 2009) I am attempting to complete our last test upgrade of a Drupal installation from version 5 to version 6. I’ve done this upgrade twice already while working on a local server at home, but this is the first time I’m trying it at work, on “real” servers. I’ll be using this blog post as a running log of the experience, with gory details after the break.Our Starting Point

The stack we’re working on is made of two servers – one for web, the other for the database.

Web:
Windows Server 2003
Apache 2.2.11 with mod_ssl
PHP 5.2.9-2

Database:
Windows Server 2003
MySQL 5.0.x (not sure the point release)

The website in question is a community resource for our school. It runs on 5.19 currently, with the abac theme and perhaps two dozen additional modules.

Hiccup #1: the system directory

The first hiccup I’ve come across is the /system directory. This is actually pre-upgrade – in order to do the test, I’ve copied the database and all files to our testing server, and added an IP-based virtual host to accomodate the test. In setting up the content on its temporary home, the user headshots have broken. They work on the production server, but on the temporary upgrade server, they’ve broken.

The files are being referenced at a path like “/system/files/pictures/picture-97.jpg” – which doesn’t exist. On both the production and testing server, the headshots are actually in the “/files/pictures/” directory – but the production server is capable of re-mapping a request to the “/system” directory, while the new testing server is not.

Core Update Complete

After following the first nine steps of the upgrade.txt instructions, I’ve now completed the core update. This has generated five messages, but no hiccups or problems. The five messages are, essentially, to:

1. Update the text of the welcome email for users awaiting administrative approval
2. Enable the update status module
3. Flesh out edit/delete permissions, as Drupal 6 has split those actions into different permissions categories
4. Look at user signatures, as the default input format has been changed
5. Adjust permissions with the Blog API module, as those permissions are distinct from blog permissions

Hiccup #2: HTTP Request Status

After completing the core upgrade, I’ve enabled the update status module and run cron. However, the status page lists another problem – “HTTP request status” fails. I wonder whether this is related to the first hiccup, that of the user profile photos not showing up correctly. I’m going to try and upgrade the first batch of modules (activitystream) before looking at this more closely, but this will have to be something I tackle sooner or later.

The first link I’m looking at to resolve this is http://drupal.org/node/222454 – not sure if it is going to be useful.

Activity Stream: file system settings

The production server’s upload directory (actually the whole drive) doesn’t exist on the temporary server, so I had to re-map file uploads to a new drive/directory. No problem there.

Just a quick blog post for now, to talk briefly about an issue we had recently with the JW FLV player. We were trying to implement poster images (those still frames you see in the player window before you hit “play”, that tease the video contents), and had a problem where the movie wouldn’t play as soon as the image parameter was added.

The implementation code we were using looked like this:

<div id="container">
  <a href="departure.asp?url=http://www.macromedia.com/go/getflashplayer">Get the Flash Player</a> to see this player.
</div>
<script type="text/javascript" src="/javascript/swfobject.js"></script>
<script type="text/javascript">
  var s1 = new SWFObject("http://support-server/swf/player.swf","ply","615","460","9","#FFFFFF");
  s1.addParam("allowfullscreen","true");
  s1.addParam("allowscriptaccess","always");
  s1.addParam("flashvars","file=http://streaming-server/2009-01-12_quale_fordownload.flv");
  s1.write("container");
</script>

(note that I’ve substituted real server names for dummy ones, but the syntax is still accurate)

Ordinarily, you enable poster movies by changing the “flashvars” parameter to include an ‘image’ value – so instead of “file=url-to-movie” you use “file=url-to-movie&image=url-to-image”

When we tried that, however, the player didn’t work. It loaded the chrome for the player, and it loaded the poster image – but you couldn’t get the player to start.

After much weeping, gnashing of teeth, and experimenting, the way we fixed it was to use a local, relative path both to the player.swf file, as well as to the poster image. The final hurdle turned out to be that we were referencing the player.swf file from a central support server, rather than that local web server – so changing this line:

var s1 = new SWFObject("http://support-server/swf/player.swf","ply","615","460","9","#FFFFFF");

to this:

var s1 = new SWFObject("/swf/player.swf","ply","615","460","9","#FFFFFF");

seems to have done the trick. We were unable to change the video location to a relative path, as that file is being served by a central University server – but making the other two files relative was good enough.

You can see the final outcome of all of this on our lecture page – we used Jason Kentner’s lecture as a test case, although we have videos in progress for most of our winter/spring lecture series, “Confluence”.

Right now, the largest project on my web development plate (as opposed to the digital fabrication plate)  is the migration of the school’s Digital Library to a new infrastructure. It is pushing me into new areas of web application development, which is both exciting and a little nerve-wracking at times. Tonight, I feel a need to write about one of the bigger areas of uncertainty.

The current digital library holds more than 35,000 images and files, and has existed in more or less the same code base for around eight years. The stack it runs on is an interesting one, incorporating ColdFusion, SQL Server, and a Scene 7 image server. I believe the web server is a Windows/IIS platform, but I can’t say that for sure.

The new digital library will obviously incorporate all 35,000 images and files, but will need to accomodate a greater emphasis on video content. This being the age of “Web 2.0″, social networking, with concepts like the mobile web and semantic web on the near horizon, however, a great deal of the framework around those images and files is being reconsidered.

Our current plan is to move to a Windows/Apache/MySQL/PHP stack, and use Drupal as the content management framework rather than a custom-built application. I feel fairly confident that these decisions give us a fairly good foundation for success, but there are looming questions. The biggest of which is:

What are we going to do for media server(s)?

As I’ve said above, our current solution uses Scene7, and the handful of videos in the collection are stored externally, on the University’s central streaming servers. However, with the migration to internal infrastructure there is a strong desire not to have the media assets themselves scattered around the University – so we are considering bringing the images and video inside the school network as well.

One of our first assumptions has been that, with 35,000+ images, our collection is large enough to demand a dedicated image server – and that with the addition of dozens of videos per year we will need a dedicated video server as well. In the back of my mind I wonder about this, but so far I’ve been content to trust that the decision to deploy an image server some years back – as part of the first digital library – is probably still the correct one today.

With the question of whether we need some sort of media server seemingly addressed, the question then becomes which server(s) we add to the project? Thus far I’ve been somewhat taken by the Djatoka image server, as well as the Red5 streaming Flash server.

Each of these platforms has their challenges, however. The most significant of which is that Djatoka seems to be dedicated to JPEG 2000 as a file format, which is barely supported by the browser ecosystem (see this 9-year conversation about Firefox and JPEG 2000). A barrier to a successful deployment of Red5, on the other hand, is that its support stack (Java/Tomcat/Red5) is one with which I am almost entirely unfamiliar.

So, this is one of the big questions that I have. I’d greatly appreciate any input from the community of people who follow these platforms more closely than I have. Stay tuned, though – I’m really excited about the direction of this project, and if we can resolve some of these issues I think the outcome will be very, very useful.

We’ve made a slight tweak to the student project gallery on our website, deploying SeaDragon to display presentation boards. You can see an example in action here:

Lindsey Warner: Courthouse Theatrics

I think you’ll agree that this is a vast improvement over our previous method of distributing boards, which can be seen here:

Erica Wicks et al: Community Plan for Eastern Harrison County

The workflow we’re using is as follows:

1. Open the presentation board (usually submitted as a PDF document) with Adobe Photoshop, rasterizing the file at about 5,000 pixels wide (around 100 Mb).
2. Save the resulting image as a TIF image, because Deep Zoom Composer can’t open vector files. If there are multiple boards, save each as a separate file.
3. Create a new project within Microsoft’s Deep Zoom Composer, and import the TIF image(s).
4. Place the TIF image(s) via the Compose panel.
5. Export the project in SeaDragon Ajax format, and copy the GeneratedImages folder to our webserver.

A few notes about this process:

Ease of use

While the process of putting together any one presentation layout is extremely easy, it will need to be faster as we scale up how we use this technology. The digital library we maintain at the school has more than 30,000 images – so doing these by hand will get old quickly. One of my next research questions will be whether we can automate any parts of the process.

My ideal solution here would be to provide a PDF document to a script, and have it follow the workflow above. Other deliverables from the script would be a thumbnail image, the requisite database entries in our CMS, and a link to the original PDF for people to download should they wish to.

SeaDragon vs. Silverlight

When I first saw this technology, I was impressed that it allowed users to intuitively explore large graphical datasets. Unfortunately, however, I can’t find good penetration data for Silverlight. The most recent references claim it is around 20-25%, but these go back to February of 2009. Hopefully with Microsoft using the technology to broadcast Wimbledon and the 2010 Olympics that figure will go higher, but until it approaches ubiquity I’d rather not fork the user experience for something this central to our traffic.

Thankfully, we can use SeaDragon – which uses AJAX (and thus Javascript) that is almost universally adopted.

The thing I’m less clear about, at this point, is how technologies like this address accessibility concerns. Surely a screen reader would not be expected to describe the various regions of a layout as they become visible, but in some ways that might be what is called for?

I’ve spent a good portion of the day today trying to duplicate – in Drupal – something that I can build in about 30 minutes by directly writing source code. At this stage, I’m going to throw this out to the Internet in the hope that someone else can help me figure out what I’m missing.

Several years ago I worked on a team that took the Herrick Building Archives online. This archive is a complete history of every structure that’s ever existed at Ohio State. The central focus of the archive is a building record – like this record for Ives Hall, former home of the Knowlton School of Architecture (where I work).

A building record is comprised of two basic pieces of content: the core building record, and zero or more addenda that were written each time the building was renovated.

To accomplish this in Drupal, I used CCK to create two content types: Building (for the core record) and Addendum (for each addendum. The Addendum content type has a node reference field back to its parent Building, which establishes the data relationship. When I coded this by hand, the addenda table had a foreign key back to the building table ID field.

The display page on the original site, then, essentially followed this structure:

1. Get Building ID from querystring
2. Look up core building record (date, text, etc) from building table; print that information
3. Look up all records from the addenda table which reference this building record (i.e. “SELECT * FROM tbladdenda WHERE BuildingID = 7″)
4. Loop through that recordset and print out the addenda text for each record

Easy, to the point of trivial. But how to do this in Drupal?

My first thought was to build a View, and take advantage of Tokens. Build a View that looks up all addenda which reference the current building ID. The problem I ran into, though (I think) is that the URL I’m working from is something like

http://mysite/building/Ives-Hall (building name)

rather than

http://mysite/building/7 (building ID)

This presented a problem when I tried to specify the argument in the view, because the preview only worked when I specified the building ID – not the building name.

My next thought was to make the entire page a view, using a Page display to show the core building record and then using an attachment display to look up the affiliated addendum records. That hasn’t worked either, perhaps for the same reason. I suppose this might be able to be fixed if I could just accept using an unfriendly URL, but I get the feeling that I’m missing something simple.

So – for those of you who read this blog – what am I missing?

I recently had the chance to pick up a used iPod Touch (thanks, Mark!) and so far it proven a welcome addition to my technology arsenal. I got it at first to replace my mp3 player (so I could give my old one to Carole), but it has quickly become more than that – a game system for our son, an offline email client for me, and a rival for my laptop when it comes to working with some portions of the Internet.

For now, though, I don’t feel the need to upgrade to a full blown iPhone. I’m happy with intermittent connectivity where I find an open wifi spot, and frankly don’t have the budget to pay a monthly fee to AT&T. Maybe in a year or two.

The real reason why I’m writing this post, though, is to experiment with the WordPress app, which I hope will prove useful for a project or two at work. We’ll see how much I end up using this app versus a web browser.

I recently became aware of the Journal of Technology, Learning and Assessment as part of a project at work on open education. After scanning the list of articles published (which appear to be all available online in PDF format!) I found one titled “Individualizing Learning Using Intelligent Technology and Universally Designed Curriculum”.

The author, Michael Abell, seems to be operating more from an undergrad or even K-12 perspective in some of his statements – compared with an undergrad/graduate design school where I work – so the paper may not be directly applicable to my project. But, he does raise some interesting questions.

One of the first topics the paper covers is the notion of student learning styles. He seems to be arguing that one of the hallmarks of a successful individualized learning system is that it would be able to configure its contents to adjust to the learning styles of any given student. Given what I remember about the different types of intelligence and the variety of ways in which a student might prefer to learn, that seems like it might be a recipe for creating about seven different versions of the same curriculum; each tailored to a specific type of learning, but each differing somewhat significantly from the others.

That seems, to put things mildly, like it would be a time-intensive process. Noble, certainly – useful almost probably – but time consuming. Given that one of the promises of open education (at least as I’ve heard it from some quarters) is that it would free up instructors to be able to pursue more research, I wonder whether those two things are mutually exclusive.

Still, I’m intrigued. Another part of my job is to train students in how they can safely operate equipment like laser cutters, CNC mills, and other digital fabrication apparatus. In that type of instruction, the “bad thing” students need to avoid is not a low mark but personal injury or damage to property.

In that sort of context, the effort required to provide seven different variations on a core curriculum is much easier to justify.

Perhaps it might be time to either schedule a consultation with some local learning science experts, or to brush up on those different types of learning so I can start radically revising our fab lab training presentations.

There are a few trends I’ve noticed within architectural education on which I’d like to ruminate – having to do with blogging in school.

One of the websites which I follow somewhat regularly is Archinect – specifically their School Blog Project. This is an initiative where they identify one or more students at each school of architecture in the country to blog about what they are doing in their school. It offers a good way to get a glimpse of what is going on at various schools around the country.

One drawback to that approach, however, is that if you get the “wrong” student from a school, you don’t get an accurate picture of what the school is like. Or, if the student doesn’t post frequently, you get no picture at all.

Another example of this sort of blogging, which gets around what is essentially a sampling error, is to have an entire studio blog about their work. I’ve seen this done at Cranbrook, Clemson, and currently at UNC Greensboro; I’m sure it has been done elsewhere as well. This seems to offer a much better impression of what a student’s experience is likely to be, simply because you get a sense of the entire studio participating.

Extrapolating from these starting points, I’ve been working over the last few months to develop a blogging platform for our school. The goal is to provide a digital commons for everyone in the school to blog about their work. We’re getting ready for the formal unveiling at the beginning of next quarter, and in the meantime I’m frantically trying to work out as many kinks as I can.

Stay tuned…

(I’ll have to edit this post tomorrow with relevant links to the various sites I’ve referenced here)

(I haven’t forgotten this profile – here’s some love, 360-ers – this is also posted on my MySpace and Facebook profiles, if you want to play along there)

The Rules:
Step 1: Put your music player on shuffle
Step 2: Post the first line from the first 50 songs that play (no matter how embarrassing)
Step 3: Strike through the song when someone guesses the artist and song title correctly
Step 4: Looking them up on Google or another search engine is CHEATING!
Step 5: If nobody gets the song right, then you may post further hints.
Step 6. Make your own

1. Yeah, hey, When you wish upon a star, dreams will take you very far

2. If I cried, if I said I’m sorry, and if I told the truth a million times

3. What a feeling, what a feeling, all alone, I had a cry

4. Well bless my soul, what’s wrong with me?

5. Another Saturday night and I ain’t got no money

6. Why don’t you move, damnit?

7. Hey baby, is that you? Wow your hair got so long

8. I could feel you across the miles in my body

9. Where oh where can my baby be? The good lord took her away from me

10. Kindness, in your eyes I guess you heard me cry

11. They’re justified and they’re ancient and they like to roam their land

12. The light, the heat upon my face, the steady weight upon your chest

13. This is HEFW waste. Sweet chocolate starfish. Read. Over.

14. There is a house down in New Orleans they call the Rising Sun

15. Your eyes are burning holes through me. I’m gasoline.

16. The night is chilly as the stars above. The things you said that day made me feel loved.

17. Should I thrist or need, can I beg you for some water?

18. Where were you when I said you remind me of a portrait in a dream?

19. My heart, my soul, my mind, I pity me for true.

20. Breakin’ my back just to know your name, seventeen tracks and I’ve had it with this game

21. I just died in your arms tonight, it must have been something you said.

22. There’s a training course where boys and girls of real ambition start a new job in a factory

23. This time its on my own, minutes from somewhere else, somewhere I made a wish

24. Danke schoen, darling danke schoen. Thank you for all the joy and pain

25. I don’t need a shrink to tell me what to think, there ain’t no missing link in my love line

26. I’ve had one cup of coffee and a cigarette, then I roll out of bed with my shirt soaking wet

27. Finally found what I was searching for, it was here all along

28. When I was a young man, I carried my pack and I lived the free life of a rover

29. I just got your message baby, it’s a sight to see you fade away

30. You had something to hide, should have hidden it, shouldn’t you?

31. I’d rather be a sparrow than a snail, yes I would

32. Thinking about all our younger years, it was only you and me

33. Wanted 19 to 5, I found it hard to describe

34. I’ll buy you a diamond ring, my friend, if it makes you feel alright

35. Every single day we follow every rule that we know

36. We all came here for a party tonight and you’re gonna get left if you don’t get right

37. Hey you’ve got a lot of nerve to show your face ’round here

38. She won’t recover from her losses, she’s not chosen this path but she watches who it crosses

39. Come and gather ’round people whereever you roam and admit that the waters around you have grown

40. Baby, I’m about a thousand miles, a thousand miles from my front door

41. I sat there looking ugly, looking ugly and mean.

42. If I had a million dollars, well I’d buy you a house

43. Another turning point, a fork stuck in the road

44. Above us lay the burdens, below us lay the truth

45. I want you to want me, I need you to need me

46. Yes this is a campaign slithered entrails in the cargo bay

47. I bought my baby a red radio, he played it all day a go-go a go-go

48. Lay down your arms and surrender to me. Or lay down your arms and love me peacefully

49. She knows that the devil’s got a burning fire inside of me

50. Clever got me this far and tricky got me in