Damn Murphy!

Murphy’s law strikes again. The moment you publish something (be it a textbook or a software), you find a bug. Yesterday I put on the web new versions of both my diffyqs notes and my real analysis notes. Already yesterday I found that when I posted the real anal notes on lulu, I forgot to update the book cover to state that I’m now at UCSD. OK, that’s a minor thing, who cares.

Today I was preparing for my diffy qs class and found an error in the notes. Actually I think I spotted this error in the spring when I taught at UIUC, but somehow forgot to fix it. So I find the first significant typo in the notes the day after posting new versions of the notes. And putting up new versions of the diffy qs notes is not as trivial as it may seem. It takes about 2 hours just to build the HTML version. This is because I have tex4ht do all math as images. Doing some of the math using CSS is faster, but then you get different font for some equations (as some are done using images and some using CSS).

If jsmath actually worked right, I would use that. But since I generally have a hard time making jsmath display things correctly on my own system, I assume that it doesn’t work right for a lot of people. Also jsmath seems to break for me for the pages as large as the notes. Mathml would be the perfect solution if it would work properly and consistently on all browsers. Right now for mathml and tex4ht you have to first decide on a browser to support, which beats the whole idea.

Browser developers are generally interested in floating 3d fish rather than actually useful stuff like supporting mathml properly. One of the reasons I am sticking to firefox is because the fact that they do support mathml reasonably, and have for a while, is an indication of sanity on the part of the developers.

number of unique ips

So I assigned homework in my differential equations class from my diffyqs notes and
the number of downloads from new ips jumped from the regular 90-100 to first week 131 and then this week (when the homework was due, 363. So subtracting 100 from each week for non ucsd downloads (I could check the logs, but I don’t have so much disposable time) I get that approximately 294 (let’s round that to 300) downloads from unique IPs are from my class. There are 180 students in my class, so each student on avarage downloads the notes from 1.67 different IPs (different devices).

On the other hand, the real analysis notes are holding at approximately 200 new ips a week. I know for sure that these are used in at least one class this fall. Given that the jump from summer to fall was from approx 110 to now 200 a week, that would over the last few weeks perhaps lead to up to 350 downloads from new ips that come from students who use that book, meaning about 200 students extrapolating from the diffy qs classes. Since the usual class size for this is 20-40 students, that mean 5 to 10 classes might be using this book. On the other hand, it may simply be students that find the book as a second reference. It would be interesting to know the real number.