11.20am: I am informed that all the relevant exam marks are in, so I make a start on producing the stats for this afternoon's meeting. I have about 2½ hours.
11.30am: user arrives with a laptop and wants various things sorting out, including fetchmail (which I actually know very little about). He's noticed that /var/spool/mail is empty, so I suggest that his mail spool will be created automatically if he mails himself something. He tries "pine", which isn't installed. I try "mail" which fails because /usr/sbin/sendmail isn't installed. He goes upstairs to get the install media for his distro, which happens to be Mandrake (something I've hardly ever used). We install mutt and postfix - pine doesn't seem to be on there at all - and it ignores the disc and downloads it off the net instead. For this we have to find a network cable and restart the network subsystem, which makes the X11 authentication stop working. And so on. This all takes far longer than it should have.
12.00pm: I am nowhere near finishing that when AWR appears and asks me how I am doing with the stats. I spend 10 minutes generating the text print-outs in front of him and print them out, and promise to have the graphs ready in time for the meeting. Now I did write (or rather, steal the code for) the graphing program, so it's my fault really that the print button doesn't really give the right result - but it's in tcl so it isn't exactly easy! So instead I investigate xwd, translate to pgm and use "tr" to remap the colours.
AWR collects the printouts and says "by the way, can you also make a chart of the following information?". "We'll see," say I. This involves writing a new program (although it turns out it is easily hacked from one of the existing ones).
1.10pm: I'm almost ready to print the graphs. AWR arrives and says he likes the printouts, but can I change the parameters thus. One of the changes sounds rather non-trivial, but it turns out not to be a change at all, thanks to some logic I added last night. Fortunately the other changes are trivial. I generate and print out the new tables and print the graphs, while attempting to eat some lunch at the same time.
1.50pm: I'm nearly on the last line of the program which was requested earlier (but it still has to have its syntax errors fixed). AWR asks me if I have his printout yet, so I tell him "in theory" it's almost ready to be generated. He says the printouts are fine, "but" can this column show Z rather than Y, although it doesn't matter for today's meeting because Z happens to equal Y at present.
2.00pm: The program actually worked! I take the printout to the meeting room and ask if they need me to sit in. AWR takes my extension number and says he will ring down some parameters. I return to my office and start editing the program to make the display of Z possible in between fielding queries from people who want to know where my colleague TB is.
2.30pm: I didn't really get started on those changes. The phone rings and AWR says he would like to be able to modify the parameters interactively, so could I bring up a laptop and a data projector. I panic TB by asking if we have any spare laptops, but fortunately we do. At that moment a technician walks past, so we ask him to take a data projector up to the meeting room.
It will save time, I reason, if I just use the laptop to log into my office machine where the database system is and run it from there. We drape cables across the room only to discover that the ethernet-dongle is broken (of course we only discover this after trying the cable in several different wall ports and then phoning down for another cable). Cups of coffee are made and drunk. I run downstairs and tar the whole thing on to a floppy, so we don't need the network any more.
Fortunately, things start getting better. The examiners can read the figures on the screen, suggest amended parameters and in a few seconds see the new results. A minor bug in the printout is even fixed (while they weren't watching (-:). In fact the meeting goes swimmingly well (apart from the minor digression into FP approximation, when the program dies while trying to compute the standard deviation of a sample of size 1 because the thing it's trying to take the square root of turned out negative). The examiners congratulate AWR and me for devising a system that is much easier to use than the old awk-scripts they used to have to contend with.
4:30pm: Meeting over, and I have the parameters which will be used to create the figures for the next meeting on Monday. I return to my office and eat the banana that I brought for my lunchtime. We try to get the network card working but still fail, so TB produces a new one. I get the data back off the laptop, which now has to be given to someone else because the one they had went dead; so I ask TB if there's another one that we can use for the meetings next week. There is, but it's Windows only. I happen to have the bootable CD I recently made of my running system, so I stick it in and test it. Of course it works, since the machine is a ThinkPad 600E very similar to our new one. We verify that it was indeed the dongle that had stopped working (the card itself is fine). I start on producing the figures that the Mathematics department needs for the maths papers that our candidates took.
6:00pm: The figures are finished and mailed.
The choir is going on a punting picnic at 7pm - a fact that I had forgotten until mid-morning at about the time that all the panic started. J is feeling icky and needs company, so I phone one of the members and tell them not to wait for me to turn up.
I eat a bag of Doritos, thus finishing my lunch.
TB is still busy with laptops. He is off to Barcelona for a Microsoft conference next week (and he was at home last week), so today the whole world wants their laptop installing. One brand new IBM ThinkPad T40 has already stopped working (we had it running for a day, then it suddenly froze, and on reset wouldn't POST - later on it would boot but only get half way through the process. Now it's pretty much dead). He made the fatal mistake of telling IBM support that he had installed Linux to dual boot with the original Windows XP system, and of course they don't support that. After some persistence, some woman at the call centre sent him the URL of a bootable diagnostic CD - but there is no response from the HTTP server, and even if there were, the machine is too dead to boot the CD anyway. An old ThinkPad 390 has just started displaying the same symptoms (hence the laptop swap mentioned earlier). So anyway, we have this Compaq Presario 2500 which TB has to finish installing before he leaves, and he can't finish without my help. In fact he can't even start without my help, because it crashes while loading the firewire module during the process of booting from the Red Hat install CD. Fortunately he's found a page on the web which tells me to use the "nofirewire" boot parameter, after which it boots and he completes the initial install. On reboot it then crashes while running kudzu, so I turn both kudzu and firewire off. Turning kudzu off means it doesn't add the CD and floppy to /etc/fstab, so I add them manually. We stick in the updates CD and set it off doing that for a while.
7:15pm: While waiting for the laptop, I finish making the stats program produce Z as well as Y (no such thing as too much information). Good job the choir group weren't still waiting for me at the Cherwell boathouse.
7:20pm: I am finally ready to go home, so I start logging out. The laptop finishes applying updates, so TB shuts it down. I am called into the setup room by a perplexed TB who has noticed that it is taking ever such a long time to unload the kernel pcmcia modules. It has in fact crashed. We reboot it and shut it down once more, just to check. It crashes again. Bah. I search the web but don't find anything. (If anyone has any suggestions...)
7:40pm: It still isn't fixed but it is definitely time to go home. TB gets a phone call from his wife wondering where he is. I say mine will be wondering where I am too!