News You Can Bruise for 2000 June 8

Get out of Mordor free! Fred was talking about a site that compared lembas to Twinkies, and I found it: Tolkien Sarcasm. Also includes such gems as misleading summaries of Tolkien's works and 10 Rejected Lord of the Rings Plot Twists: "Balin emerges from the depths of Moria, claiming he 'fell asleep in the tub'."

Strangely enough, string comparison seems to already work. I'm not complaining.

Roger Ebert thinks he's Dave Barry.

Test cases that fail: 8 (integer division is done as real), 9 (similar cause), 10 (weird constant thing), 11-13 (division again!!) 15-17 (operations on whole arrays), 19 (integer division gives same results as real division (bleah!!!)), 20 (nonexistant label). That's not good (there are 22 test cases, which means I correctly compile only half of them). I really have to fix the division. The weird thing is that the TA says we don't have to implement division, yet it's used in 9 of the sample test cases. If I fix the division I suppose I can call it quits.

8, 9, and 11 work now. I have to go to class.

I'm not going to fix #19 because I object to its semantics (also, it's too hard). I fixed the conversion problem, but I'm not going to make the type of a division expression depend on what type the result is being assigned to rather than on what type is being divided by what type. I mean, come on.

#10 also looks hopeless because jasmin doesn't seem to want to accept negative values for its constants.

I'm working on the whole-array operations now. Then I'm going to quit.

I have a compiler submission in. I spent all this time getting array operations and printing out of whole arrays to work, and only one of the three test cases that uses that now works correctly (the other two have subtle things wrong with them). Bleah. I've been up since midnight. Double bleah. I have 16/22 right now, which isn't bad. I just realized that the lecture I'm going to go to right now is my last lecture ever. Yay!

I don't know what the people in the other building are doing, but I'm fairly certain that they could do it at about 3% of their current shouting level. Unless, of course, what they're doing is shouting.

17 works now. 15 still doesn't work and isn't going to work because it assumes that arrays are stored in column major order and I store them in row major order. Therefore, the test cases that have stymied my compiler are 10, 12, 13, 15, and 20. I'm working on fixing 10, but it will be ugly.

The great thing about having two techno albums is the variety! One techno album might get monotonous after a while, but when you have two, one or the other is always appropriate!

The compiler takes forever to make. I blame the Internet. Specifically, I blame HTTP v1.1.

I fixed the problem with 10 but it shamelessly manifested another problem, so I give up. 17 out of 22 isn't bad. I'm testing the compiler for turnin now.