Tell HN: My horror story as a student dealing with my University computer lab

The strange thing, to me, is that I seem to hear this a lot from people I know who go to "top notch" CS schools. Cambridge, Austin, these are supposed to be great schools!

I went to Southern CT State University, which is basically a glorified community college. ("I went to college in New Haven." eyebrows "Oh, Yale, huh?" "No, the New Haven college you HAVENT heard of.") I left in 2002, so maybe things have changed, of course.

For a college that felt a lot like a high school with dorms, we had fantastic computer labs.

In two of the dorm buildings, there were labs full of Macs and PCs (20 of each in each lab). There were another 3 PC labs in the science building, and another PC lab and Mac lab in the library. In the CS department, there were 2 unix labs, one with about 20 BSD+KDE machines, and the other with 10 SunOS+OpenWindows. There was also the VAX/VMS system that anyone could use from anywhere via telnet, and provided access to all university services (including the ability to ssh into the Unix mainframe.)

Though I didn't really appreciate it at the time, the computer science education I got there was fantastic. I've talked to friends and coworkers of mine who have masters degrees from "good" schools and I'm aghast at the gaping holes in the curricula. Many of them never had to compile a program until they entered the work force, had never seen O() notation, and so on.

I think the moral of the story is that you should be wary of judging books by their covers: the campus might be beautiful, and the reputation might be sparkling, but take a look inside the buildings before choosing a school.

> I went to Southern CT State University, which is basically a glorified community college. ("I went to college in New Haven." eyebrows "Oh, Yale, huh?" "No, the New Haven college you HAVENT heard of.")

You do not really appreciate the quality of universities in the USA. Even the university that you view as "second tier" is excellent by world standards.

Also - smaller universities that focus on undergraduate education often provide a better education than large research oriented universities.

Yes, it often seems that SCSU's only failing as a university was that the landscaping and architecture weren't as nice. The bureaucratic hassles were often pretty annoying, and from what I've heard, they aren't so bad at private schools.

But yeah, there are lots of very good, very cheap schools in this country. I have almost no student loan debt to speak of, and I got a great education. Not to mention that you can watch a lot of lectures on youtube and whatnot.

The main advantage of a prestigious school seems to be that you get to rub elbows with rich people's kids.

> The main advantage of a prestigious school seems to be that you get to rub elbows with rich people's kids.

Another advantage is that you can casually mention it repeatedly on HN (casual university name dropping).

And be completely ignored by the non-US people who're more interested in clue level than an atlantic-away university.

And the non-university people.

And all the people who ignored nicknames changing colour.

But I'm sure it'll impress everybody else :)

Another thing that I have noticed is that, the better the school's reputation, the more likely it is that people will try to BS their way through, knowing the that name on the paper will be enough to get them where they want to go. A lot of students at "second tier" schools recognize that they need to be on top of their shit because nobody's just going to give them the benefit of the doubt.

I assume you have a student union at your university? Some CS professors? Other people who know what's going on?

Get together, schedule a meeting with whoever has the power do put a stamp on whatever draft you can write before the meeting. Make sure there are some admins and the head CS department in the meeting too. I bet it's a nice feeling to write here about the situation, but HN can't do much to help you. If you see the problem - go now - book a meeting room if you have them. If it's that bad, do you have anything to lose?

That makes perfect sense, thanks. There are a few reasons why I posted it to HN:

* I had to pick my battles when I was there and this wasn't one of them. I've emailed this to friends who are current students; hopefully something will come of that.

* I wanted to confirm that this wasn't just a UT CS thing. As I expected, there are a bunch of responses attesting to similar experiences at other places in the comments below.

* I was also curious to see the discussion on what people think would be a good way to run the labs (or at all.)

For anyone who wants to take the lead in trying to get things to change at their univs, I hope this post will serve as a confirmation that you're not alone and you should be able to find many other students to join you in pushing for change.

OK, now for the contrarian opinion: you're barking up the wrong tree.

A good university does not coddle up students like kindergartners, feeding them pap. It's more like boot camp - in the mud, on foot, carrying a heavy pack. Apparently, it's pointless, sadistic, exercise: the 'product' of the work has no value. Except for the connections your synapses make ...

Frankly, it's hard to feel sorry for students who will not 'read the FAQ', or do anything to improve their conditions by themselves, if they feel it's important. It either is not important, or the culture of helplessness has reached too far.

Also: C.S. is not about this week's 'state-of-the-art' tools - as the late Dijkstra (who taught at U.T.Austin) put it, it's like astronomy and telescopes.

P.S.: in any organization with more than a few dozen people, support staff will ask for emails or web-tickets. That's for two reasons: accountability and knowledge sharing.

Agree! My C.S. undergrad program did not even cover specific languages. There were no courses such as "The 'Foo trendy du-jour' Programming Language" -- the language was incidental to the subject matter, and you were expected to learn the language, editor, OS utilities, compilers, debuggers, etc. on your own time.

There's a difference between expecting students to overcome adversity in a tough curriculum and making things pointlessly painful.

CS is not about state-of-the art tools, but it is most definitely about good tools. In fact, I'd argue that much of undergraduate CS is about learning how to recognize, use, and build good tools, and part of the way that you do this is exposure. I think that people still forget that CS is, at least partially, a creative discipline, and the solutions that its practitioners come up with are often influenced by the tools that they use.

First of all, root access to the machine should be granted to anyone at a terminal. Physical access bypasses any sort of realistic security placed on the system.

At MIT, the computing infrastructure is largely a collaboration between students and admins. This has worked incredibly well, and led us to have a continually evolving system and created a great system for tech support.

Right now we're in the process of testing Athena 10, a ubuntu-based overhaul to our old redhat system. It's a huge task, and a majority of the bugfixes and patches have come from student hackers.

You're running linux, so that's a start. I'd suggest seeing if you can get some of the defaults changed (such as fvwm2), and maybe start a discussion about moving to a more user-friendly distribution.

#

Here's a few resources about Athena@MIT: http://web.mit.edu/ist/isnews/v24/n02/240203.html http://blog.spang.cc/2008/01/10/mit-athena-not-dead-yet

My university copied one of MIT's old versions of Athena, and called it project Vincent. It was fantastic, so far ahead of anything else I'd seen in 1998. Too bad they started to phase it out and lessen support in favor of a big microsoft network. Though it happened when non technical students started to use computers too.

Iowa State University, right? I'm a student there right now. They've improved since the Microsoft dark age, moving their server infrastructure to Linux and clearing away the legacy cruft except in the humanities departments. No, the big problem now is that the computers are so locked down you can't install software on them! This isn't so much an issue on Linux or MacOS -- you can install everything to a folder easily -- but the Windows machines are a pain in the bum when you need some software you don't have. Even if you have standalone exe files they sometimes refuse to run for permissions reasons.

This is a major software issue. In the happy land of Ubuntu, I can just use "./configure --prefix=~/myroot", and everything will work out more-or-less fine. On MacOS I'm just dragging around application bundles. But on Windows this whole process is colossally painful, and the situation would probably be dramatically improved if you gave everybody the damn Administrator password and carte blanche to use it responsibly.

I'm at a university right now in a program that involves some 3D modeling and rendering. Our 'render farm' was purchased about 4 years ago, and is now barely capable of rendering the scenes the students create in less than a few weeks. So the IT guys had to come up with a solution...

They installed some sort of batch render daemon on the lab computers (windows) causing them to take, and I kid you not, 1 hour to log in. We essentially don't have a lab anymore. Your experience might make the computers seem unusable, but mine truly are for a while.

That's really sad. I'm in Russia, I thought that worst computer labs are here (CRT monitors, Windows 98/2000, 100Mb quotas, no running overnight jobs, etc).

IMHO what you're describing is a clear manifestation of oldskool 'computers are complex systems and are to be used by professionals' ideology (as opposed to 'computers are easy tools for everyone').

I also use a text login, fvwm2, and install things from source (albiet in Gentoo rather than Ubuntu). As a 19 year old kid, should I feel hard done by? :)

Maybe it's different in England, but I've found my university's computer labs (Uni. of Warwick) to be useful, and not insane. I know from experience that if I didn't have my own computer (as was the case when it was broken during my first week), I could get by without too much frustration using the university systems. I can even use dvorak.

(Noone else seems to have problems with them either, so it's not just that the admins are as insane as I am.)

Not that this is meaningful data. But I'm not sure your experiences are either. Whether computer labs provide value depends on the labs and the students. There's no answer that will apply to all universities.

No problems at back at my undergrad at Bath Uni either. You could have graduated using just the Uni machines. Plenty of desktops spread around campus, good wifi accessible with a username and password.

Plain old Fedora at the The University of Edinburgh here. Pretty reasonable setup, a few things a bit out of date, default quota a bit on the tight side if you need to install your own version of anything large (like wxwindows etc) but all in all pretty fair.

> I also use a text login, fvwm2, and install things from source (albiet in Gentoo rather than Ubuntu). As a 19 year old kid, should I feel hard done by? :)

You're accustomed to using a command line, you probably chose fvwm2 because you were looking for a no-frills window manager, and Gentoo's emerge system makes basic installation from source as easy as using apt-get. People who aren't as nuts as you and I have every right to be ticked off at computer labs that act like your computer rather than computers for mortals.

Slightly off-topic: I consider applying at Warwick for a PhD position. Could we get in contact to talk about your university? You'll find my email address in my profile. Please drop me a line. Thank you!

I grew up at a time when unix computers were UNIX computers, SUNs and HPUXs. The HPUXs were student run. They all had interfaces similar to fvwm2, and I don't recall having many problems learning how to use it. In fact, the reason that I can navigate so well in linux is because of the things I learned from doing stuff on the unix machines back in college.

There's nothing inherently wrong with fvwm2. I'm sure it was cutting edge back in your day. But it's ridiculously outdated in 2009. You might have learnt to use it, but today's students need to unlearn everything they've learnt. Even if they manage that, it's plain inefficient. By a big margin.

Whatever technology you grew up with, you're bound to think of that as the natural interface. That's exactly the point I was making about the professors.

There was a time when you had to do typesetting by typewriter. There was a time when all cars were stick-shift. There was a time when farmers had to plant seeds manually. That doesn't mean it's a good idea to do any of those things today.

I'm not sure if you were claiming you will learn linux better if you use fvwm2, but that's simply not true.

But that is not the main point I'm making. What you need to understand is that no matter what you, the admins or anyone else might think of fvwm2, it's turning away the students in droves. This needs to be stopped.

> fvwm2, it's turning away the students in droves. This needs to be stopped.

Why ? Is the school losing any good computer scientists ?

I had to use the computer lab for many of my comp sci courses. If the lab had had an unusable window manager, then I wouldn't have taken nearly as many comp sci classes. Undergrad CS course aren't just for CS majors.

I think uni labs should just focus on providing great hub access - big monitors, ethernet, wifi, disk. Students should be issued or be assumed to have laptops and there ought to be a tech support center. Why do colleges continue to overbuy hardware when it's clear from the outset every year how many machines and licenses are needed?

Its hard to imagine a large group of CS majors managing to maintain a Unix desktop suitable for development when pre-class discussions in senior-level classes involve people complaining about the difficulties involved in compiling from source.

I'm pretty sure most of the money saved by ditching the CS labs would go towards Unix classes and a 24/7 support hotline.

In my experience, the hacker qualities of CS undergrads fall on a pretty broad spectrum. Even if the majority of students have trouble with basic Unix tasks, I'm sure there will be 5% of them who will make excellent admins. And that's more than you need.

So how do you propose we set things up? Most of the need for CS labs is conformity in grading/teaching. I see the appeal - both fiscal and hacker-practical - for dropping the Uni labs, but I think that forcing 95% of CS majors to rely on 5% of their peers for support would cause enough grading and deadline ("I couldn't get the package to compile in time!") arguments that both faculty and undergrads would come to hate things even more.

Granted, there are obvious upsides to forcing people to learn systems other than Windows for their own use, and I think enough people would eventually adapt, but I think the initial hurdle would just be too much to overcome in a decently sized department.

In college I often sshed from my computer into a lab computer, so I didn't really need the physical lab. Of course, maintaining servers may cost almost as much as maintaining a lab. Maybe preconfigured VM images are the answer.

People ssh'd into a computer are seldom using all of its power. A big multicore server with plenty of RAM can support quite a few users more economically than individual computers. This is part of the reason why people invented timesharing OSes in the first place.

I'm sorry, I thought that the CS computer lab was for doing CS work.

Yes, there's some stuff you probably should have had available -- automount and a better print manager come to mind. But, really, what are you doing in CS classes that can't be done well using fvwm?

I graduated from UT with a degree in CS a couple of years ago. Those Dell machines are God-sends compared the the Solaris ones they have. Last time I tried to use the Solaris machine it didn't have ls installed on it.

I totally agree with you about the Linux setup. The first time I start fvwm2 I was like wtf is this. That actually encouraged me to figure out what a window manager was and how to start a different one.

You are wrong about the overnight jobs (at least when I was there), you can only run them on certain servers that are designated for overnight jobs. For most cases you did not have to get permission.

UT has enough money and computers that each student should be given their own vm.

Why would anyone use the computer lab? How many undergrads actually do any computing for which $500 desktop computers in their dorm rooms would not be more than sufficient? Do the labs require special, non-OS software? In this day in age, that would be really weird. Even if classes are taught using Microsoft stuff, the basic tools are all free anymore.

I recall reading that universities are doing away with dorm phone lines -- between cell phones and VOIP, students have no use for them anymore.

This links in a lot with the other computer lab thread, but for some CS teaching the software setup is particularly arcane, and classes are taught in the labs themselves, using that setup. You can't really avoid them, setting it all up on your own machine, learning to use it without the practicals and so on is just pointless extra work.

The basic tools may be free, but the more advanced things aren't. Many assignments require the use of packages like Matlab, Maple, and Pro/E - the idea being that we'll use these things in industry so we may as well get used to them in university. There are also a series of packages that are used that are free, but hard to find/maintain individually.

It also comes down to standardization - if I'm marking circuit layouts, I want everyone to use the same version of Electric that I did, so it's easy for me to deal with fifty submissions. I want everyone to compile their code using the same environment, so that I can simply recreate that environment and know that the code should compile.

For the social aspect.

There was a famous story about some University's computer lab (I forget the details) where they got fed up of the coffee machine. There was a perfectly good cafeteria close by where the students could get coffee. Plus they had to keep the machine stocked, service it, etc. It was just a pain in the ass, so they got rid of it.

Within about a week, their TAs were swamped with demands on their time from students. They had to take on like 2 extra people to meet the demand.

Turned out, that coffee machine was a social hub, and people would chat about their problems, ask each other for tips and advice, and solve a large percentage of their problems on their own.

So they bought a replacement coffee machine.

Now I'm imagining those students doing all the work on their computers in their dorm rooms on their own. Hmm.

Sure, there's lots more connectivity these days and places you can ask for help, but when learning/starting out that sort of personal contact etc is quite useful.

Sounds like a case of the bad ol' "I'm the boss here!"-style 'administrating', eg. some grizzled unix wizard knows only HIS way of doing things, that's automatically the way things SHOULD be done, these young whippersnappers'n their fancy GUI-driven installers, get off my lawn etc.

All they have to show for themselves is a flawlessly running system, which, to Those That Make The Decisions, is all that counts.

It's too bad to hear this.

I am very satisfied with our computer labs. We're having 7 labs with about 10 PCs each, all of them equipped with recent Intel CPUs and quite a lot of RAM. Windows or Linux (Debian or SuSE I think) on all machines. Possibly all apps that you're ever gonna need installed (ranging from python+IDEs to MS Visual Studio, Qt, Gimp, MS Office, Open Office, ...). No hard disk quotas are enforced, instead there's a fair use policy. Admins are very friendly and talkable to about anything...

And if you don't like the machines in the lab, grab an ethernet cable or use the WiFi available all over the campus.

The labs are used by many students, even though I'm not aware of anyone without a laptop or a machine at home. It's just too comfortable... sit down, have all the apps, all compilers you need for your studies ready for use.

In my opinion, this is how you should run a computer lab - user friendly policies that encourage the use of the public machines. After all that helps the students to get to know each other better and help each other.

For the record, I study at Hasso-Plattner-Institute in Potsdam, Germany.

Drexel University had decent labs of every variety: Linux, Mac and Windows. The Linux labs -- used by the computer science and software engineering students -- had reasonably modern single-core machines with 4:3 LCDs approximately 20 inches in size. There was often a shortage of available machines around project deadlines. However, the digital media students, just one floor below us (in the basement), had beefy Windows machines with larger dual monitors and access to powerful render farms! They had plenty of machines to spare, but CS and SE students couldn't even get onto that floor with our key card access. I have no idea who had access to the Mac labs. It certainly wasn't us or the digital media students.

I'm pretty sure the reasoning for this is as follows:

1) Computer science and software engineering students prefer to use their own computers

2) Digital media students can't afford the expensive licenses for the software they need to use (Maya, Max, Photoshop, etc)

3) Typical personal laptops aren't powerful enough to do proper rendering work

4) You can always SSH in, use VIM, and compile from your personal laptop

Back when I went to school (mid-90s), our computer lab was a bunch of orange-screen vt220s connected to a Vax cluster. We didn't have fancy NCSA Mosaic like UT, we had Lynx and gopher. We learned C on a VAX with a broken scanf. Students dropped CS because they couldn't figure out EVE and EDT (visual editors). Our internet connection came through a narrow tenuous pipe from UT Austin. When it went down, I'd do a traceroute and then give "Mordac" over there at UT a hard time.

Oh yeah, there was a PC lab. Running Windows 3.1. That was a place to type your paper. You couldn't do real CS there.

And I had to bike uphill to class both ways in the Texas heat. :)

I think if you can take busted up hand-me-downs and backwards old crap, make it work & explain how to others, and learn how to sidestep a lazy/stubborn/indifferent IT (build your own stuff in ~/local, etc.), you can adapt and do fine anywhere.

I went to UT cs also.

While its not ideal, I thought the cs computer lab was fine. It's not meant to replace your main computer, as a CS major you should probably have your own computer. Its meant more as a place to hang out and work on class projects together and for that I think it does fine.

Also, if you are a cs major you can probably figure out how to 'startx gnome'

Also, for many research groups they have their own computer labs that are not restricted like the main cs lab. I know at least that in the robotics lab they gave out root passwords to their ubuntu machines and you could pretty much do as you like.

> Also, if you are a cs major you can probably figure out how to 'startx gnome'

Don't imagine yourself at a computer, think about the marginal user. That is, the kid who could be a good CS student but doesn't do it because the first class' experience is extremely frustrating not due to programming, but due to stupid computer admin shit.

The argument is that the unnecessary complexity and intentional backwardness of the computer labs is pushing away marginal CS students, not that it's a system that's impossible for anyone to figure out.

A tweak on the "let the students run it" idea would be, for undergraduates at least: Let each year's class run their own systems. I could list a bunch of benefits (and a few drawbacks) that would fall out from this, but they seem pretty clear, probably no need to list them - they range from the social to the technical to the logistical.

Another tweak: keep a locked-down simplified version of the operating system on a separate partition. Post signs telling the newbies how to get to it if the main OS got broken by some student. Hard drive space is cheap, right?

Well, for what it's worth, at Stanford every cluster I've used works just fine. I wouldn't mind a bigger quota (I don't think I even get 1GB) but other than that, no complaints.

Well, it would be nice (no pun intended...) to force people to run long MATLAB jobs with nice, but oh well.

I see nothing has changed since I was there ten years ago. At that time the CS admins were students, but the job seemed to attract the BOFHest of the BOFHs.

If you had an office you probably should have just put your machine on csres and run it yourself.

USC had a couple big (by 1990's standards) time sharing Sun machines we used for CS and EE classes, which you could login to via SSH or VNC. Or you could use one of the few dozen Sun workstations in the lab which had the same software but ran locally, I guess.

The window manager was CDE by default, which sounds about as bad as fvwm2. You could run Gnome but I don't think most of the software worked very well with it (ePD! Cadence!)

Fortunately most CS projects could be developed locally on any Unix/Linux/OSX machine, then thrown up on the time sharing machine for testing.

Northeastern's CCIS computer lab is a little bit like that. Lots of rows of SUN workstations with some crufty old default window manager (even though you can switch to Gnome).

I find myself doubting the veracity of this article, because I just cannot imagine this kind if lunacy. I majored in physics at the Delft University of Technology and although it wasn't perfect, I might as well call it perfect compared to what you are describing. The helpfulness of the IT department in attaining goals made up for every shortcoming.

I went there, its true

I was about to go to UT Austin, but after this story, I'm glad I came to Georgia Tech instead. The lab computers are pretty good, and the ones in the College of Computing are just thin clients that VNC you into Red Hat or Windows, which you can VNC into from your own laptop. On top of that, the Wifi network is fantastic.

At UIC they have two kinds of computers. Computers that run according to ACCC specifications and computers that are owned by the department of computer science. Guess which ones are easier to deal with...

s/University of Texas at Austin/University of Cambridge/

All true, up to the point about package management, which didn't quite take a week.

But as a result I use fvwm2 as my coding WM of choice. It's distraction-free!

hey, fvwm with the right configuration is quite nice... It was my windows manager of choice before switching to os x 4 years ago (and I still miss it some times).

I used to run the gnome panel with it and found it much nicer than metacity+gnome...

But of course, fvwm is good only if well configured and I doubt they took the time to do that. And I do agree that most people will not see want to spend hours configuring it...

But the uni is for CS right? I think your uni has made a lot of good hackers :).

Berkeley's Air Bears Wi-Fi works pretty well most of the time.

At UT, perhaps it's a Dell plot to defend Wintel.

You make some excellent points. My experience have been with Windows. The situation is this – they install shitloads of programs on these machines (most of it BS) . This all run at startup – the system tray is half the taskbar.

All this crap makes the machine maddingly slow. Then they mount network drives on the machines (20 or so). This means that if you use a flash drive you must start unmounting the network drives until your flash drive appear.

While you log-in 10 or so programs give you error messages (for which you must press the Ok button).

The quotas are still a problem – why must we have 100MB quotas? Hard-disk space is cheap – google gives me 7GB for email.

The worst is the internet costs. They give you 200MB for a year. After that you have to pay R2 (20 US cents) per M.B. That price is four times as expensive as GPRS through your cellphone.

Oh – it is crap working on CRT screens with a burned in log-in bar.

---

Ranting aside – the admin for the clusters in my dept. really knows his stuff.

Yeah the quotas are ridiculous. I don't think we had a bandwidth quota in the labs though. 200MB seems pretty extreme. In the dorms the quota was like 10GB a day.

I just logged into my account (I graduated more than a year ago but I guess they forgot to deactivate my account...) and it looks like they bumped the storage quota from 100MB to 200MB. Yay.

Let the students run their own machines?

Considering the difficulty of the entry-level classes (weed out) and the fact that students often have 3-4 extra classes just like it, having them spend hours upon hours of admin work. They're training to be computer scientists not admins.

My school runs RHEL/gnome on some beefy quad-core machines. Whatever is needed for a class is installed immediately and just works. No need to spend hours hunting down packages and installing them, and then no need for the professors/TAs to spend hours upon hours of duplicating configurations in order to make sure that everything can be graded.

Students will acquire enough Unix knowledge throughout their 4 years by using a well-maintained system. Most of them spend about 10-20 hrs/week using it for development -- no need to add another 30 for sysadmin work.