in reply to Seeking advice to fix an app that I have no control over

E-Bitch,

I used to administer a WebCT installation. At the time it was one of the biggest in the world. (1,200+ course accounts and 30,000+ user accounts.) Later, I worked for WebCT proper.

Down almost constantly.

WebCT can be rock solid, but we ran it on very beefy Sun hardware and Solaris as the OS. It was bullet proof. Our only down times were network issues; cut fibers mostly, once on campus, once in Reseach Park adjoining campus, and once in downtown Orlando. If you were on campus, you could still get to the server. UPDATE: Is Colorado State running WebCT on Windows?(I hate to sound like I'm bashing Windows, but WebCT is more stable on *nix...) I just found out that CSU is indeed running WebCT on Unix...

You'd be suprised to hear that many institutions consider WebCT mission critical. Often times, it is understood to be just as import as payroll systems and campus wide email. And that makes sense - if your school is offering courses with required on-line components and those components aren't available to the students, that's the equivalent of the school not being able to hold classes in classrooms due to maintence issues (no blackboards, no desks/chairs, no running water in classroom buildings, no HVAC, no restrooms, etc)

For what it's worth, the WebCT admins at CSU (Jamie Bethel and Craig Spooner) are both underpaid and overworked. They both have a good reputation in the on-line WebCT community.

...looks like it was intended for one class, and was bodgered and grew into what it is today.

Kudos to you on this point, you are absolutely correct. You hit the nail on the head here. Murray Goldberg developed it to augment the computer science courses he was teaching in 1995 / 1996. He was doing the right thing at the right time in the right way and it took off like gang busters - literally.

I know it uses a database on the backend, but I've never seen it, nor know what the schema looks like.

No, it doesn't. You'd like to thing it does, but it doesn't. Actually, you're correct if you call a collection of flat files and DBM's a database. You'd have to look at the way WebCT does things to understand what I'm saying. There really isn't a schema. WebCT in current production versions uses a database like a combination of the Yellow Pages from the phone book and your mother's recipes can be thought of as a database. WebCT's "database" is certainly not object oriented, just relational enough to do what it needs to do, and any sort of DB management system is integral to the programming and not a separate system. If you still want to call it a database, it is by no means normalized.

I said Murray was doing it the right way, because in 1996, no higher education administrator in thier right mind would have paid for a DBA to run a database for on-line learning. It just wouldn't have happened. The key to WebCT's success (and market share) was that it was easy to install out of the box onto a machine of modest proportions by someone with modest technical skills. Often times, WebCT was installed by a faculty member on "spare" hardware without the knowledge of the administration or networking/computer systems staff.

... scored 100%, and the system lost the record.

As for your quiz problems, I've heard of WebCT "losing my score" and "changing my answers" on and off for four years. Trying to trouble shoot, I've taken quizzes myself and watched affected students take (and re-take) quizzes. I've watched them in my office using my machine, in the computer labs, and even at their homes on their own PC. I even had a cadre of part-time student employees take a bunch of quizzes looking for this problem. I'm not saying it doesn't happen, I'm saying I've never seen it happen nor have I ever seen anyone who had it happen to them recreate it, and I've tried pretty hard. In all fairness, there is an issue of resizing browsers with Netscape, but at best that will affect a single answer. If you can figure it out, please let me know.

If your really interested in WebCT, webct-admin-request@lists.sourceforge.net (subscribe at https://lists.sourceforge.net/lists/listinfo/webct-admin) is a good list as is webct-users-digest@webct.com (Majordomo). The WebCT SourceForge site will be pretty much useless to a student. Most of the scripts there are for versions 1.x and 2.x; although there is some good information about adding FrontPage Extension to a WebCT server. I'm guessing CSU is running 3.x.

BTW, the last I heard, the next version of WebCT (a.k.a. Colbalt/Vista/4.0) is being re-written from the ground up. Rather than Perl, it'll be J2EE Java serlets running on top of Oracle 9i.

Cheers!

Brent

-- Yeah, I'm a Delt.

  • Comment on Re: Seeking advice to fix an app that I have no control over

Replies are listed 'Best First'.
Re (tilly) 2: Seeking advice to fix an app that I have no control over
by tilly (Archbishop) on Feb 25, 2002 at 00:24 UTC
    With regards to the "lost quiz" question, a good approach would be to put a logging proxy in front of the server. Then whenever a person comes along, you can ask for their ID, then pull up a record of every page request and give them a complete reconstruction of their session, completely independently from the implementation of WebCT.

    There are some serious privacy issues. But on the flip side, having a student claim to have test results that they don't have is fraudulent...

[reply]
      WebCT allows for this kind of tracking. (Although not independent of WebCT as you suggest.) Instructors can view a quiz session history from initial load to final submission for grading, including each (re-)submission of an individual answer (each question is it's own HTML form...), each with a time stamp. It creates a nice chronological view of a quiz attempt. Looking at this history, you can even tell if a student likes to answer all the question on the quiz and then go back and correct their possible mistakes; or work on each question one at a time until they think they've got it right before proceding to the next question. You could even tell that some students would print the quiz, work it out on paper, and then "batch input" the answers back into WebCT.

      On a side note, WebCT will also calculate the upper quartile, lower quartile, discrimination, standard deviation, and mean for questions. Hopefully instructors use this information to tell when they've got a bum question, but I suspect few know this type of analysis is available to them.

      Additionally, the instructor can view individual student statistics for page views of course content, a chronological list of pages hit in WebCT, length of time spent on pages, messages read in the bulletin board system.

      Furthermore, WebCT collates all that per student data and will provide averages across all students. It even provides for longitudinal analysis across semesters.

      I agree whole heartedly about the privacy issues. I think most students would freak if they saw the depth of statistics WebCT tracks.

      If I ever had a wiff of something not jiving with the data I had available, I would have hit up our networking staff for something like you suggest to get data from outside of WebCT. (BTW, the network people at my institution were great! Their attitude was "We're here to help you do what you want to do." When was the last time you've heard networking / systems admin. people say that?) It was a pleasure to work with them.

      Cheers!

      Brent

      -- Yeah, I'm a Delt.

[reply]
Re: Re: Seeking advice to fix an app that I have no control over
by E-Bitch (Pilgrim) on Feb 25, 2002 at 00:10 UTC
    Well, I stand corrected. I naively assumed that it was running on a database. But good to hear that it will port over to Oracle 9i eventually. Unfortunately, the timeline that this will occur on is outside of my timeline at this university. And I'm wondering why the switch to J2EE? CGIs can be just as good (if not better) at the same task, and since they have already started with them... wouldnt it be prudent to continue?

    Thanks for all the help / Advice!
    _________________________________________
    E-Bitch
    Tempora Mutantur Nos et Mutamur in Illis
    "The Times are Changed Even as We are Changed in Them"
[reply]

      And I'm wondering why the switch to J2EE?

      I don't know the answer to that. If I had to take a guess, I'd say it's so that they do a true rewrite of the code base. Switching languages forces that. Staying with Perl, someone might try to reuse code and there's lot's of Perl 4-isms and such that are best avoided...

      ... outside of my timeline ...

      From what I gather, the new WebCT is in internal Alpha right now. I'm guessing that'll be followed by a limited Beta and then a general Beta... But then again, at one point I was told the new version would be out by Q1/Q2 of 2001 ... :-) (Standard disclaimers apply - I don't work for the company etc. etc. - if you realllly want something beyond educated guesses, contact an account rep.)

      Cheers!

      Brent

      -- Yeah, I'm a Delt.

[reply]

In Section Meditations