Best Abuse of the Periodic Table: John Dalbec 20 Squire's Ct. Canfield OH 44406 USA http://cc.ysu.edu/~jpdalbec/ Judges' Comments: To build: make jdalbec Try: ./jdalbec 1 ./jdalbec 11 ./jdalbec 1 1 We still aren't sure whether or not this is a useful program, but it's the first atomic fission we've seen in the IOCCC. Selected Author's Comments: It's elementary! See http://mathworld.wolfram.com/LookandSaySequence.html, http://mathworld.wolfram.com/CosmologicalTheorem.html, and the references listed there. Go read the web pages and come back. This program implements the look-and-say sequence with the added feature that each "decay product" is automatically split into its constituent "elements" separated by spaces. Furthermore, arguments passed to the program are treated as separate elements for the purpose of computing further decay products. (Compare the output of "prog 1", "prog 11", and "prog 1 1".) This allows output lines from the program to be passed back to the program as input (minus the initial generation number). The common elements are listed in CosmologicalTheorem.html. The non-primordial transuranic elements are the isotopes of Neptunium and Plutonium. Isotopes of a given element differ only in the last digit. Four common elements could also be written as isotopes of these two elements: He=Np(2), Li=Pu(2), Ta=Np(3), W=Pu(3). To get other isotopes, just change the last digit to a number greater than 3. The build file runs prog.c through the C preprocessor to generate prog2.c which is edited with ed to correct spacing problems caused by some preprocessors. (I know you dislike Perl, but hopefully ed is more common on POSIX systems than Perl.) The build file then compiles prog2.c and removes the prog2.c file. I don't think I'm abusing the build file since I'm only using 147 characters including white space and the #! line and the commented-out "-Wall -pedantic" options. The rules for the 16th IOCCC rules specifically allowed build files up to 521 characters long so my build file would have been well within those limits. (What typo?) gcc -E -Wall -pedantic complains that trigraphs are present and that "pasting ... does not yield a valid processing token". gcc -ansi -Wall -pedantic complains about #line directives, trigraphs, and expressions and statements that have no effect. The last complaint is false in all but one case. Removing the offending expressions causes some compilers to report syntax errors because they can't determine whether a name refers to a variable or a type. Since buffer overflows are a common security problem these days I have stolen the idea of using a "safe" buffer API from vsftpd. This means the program should never crash, right? The program was tested on: AMD K6/MMX 266MHz, Debian GNU/Linux testing "Sarge", gcc 3.3.2 Intel 486DX 33MHz, Red Hat Linux 7.3, gcc 2.96 PowerPC Macintosh G4 Quicksilver, Mac OS X 10.3.2, gcc 3.3 PowerPC Macintosh G4 Quicksilver, Mac OS X 10.1.5, cc 2.95.2 PowerPC Macintosh Performa 5200CD, Mac OS 8.6, CodeWarrior 11 Sun SPARC Ultra 10, Solaris 8, gcc 3.2 Compiling on Mac OS 8.6 requires some degree of manual intervention because Mac OS 8.6 doesn't have shell scripting or even a normal command prompt. The CodeWarrior 11 compiler is the reason I had to use trigraphs. I have to preprocess prog.c without trigraphs, save the result as prog2.c, then compile prog2.c with trigraphs to get the program to work. I also have to add an include file and a function call to the program to get a pseudo command line so I can pass arguments. The compiler also requires function prototypes by default, which is why I have prototypes for all my functions. Compiling on Mac OS X 10.1.5 requires changing ??= to \# on line 3 because the compiler doesn't recognize ??= as a trigraph (which I consider a compiler bug). The initial preprocessing step on Mac OS X 10.1.5 generates lots of errors in "smart mode" because the "smart" preprocessor expects a valid C program as input. The preprocessor then falls back to "basic mode" and preprocesses prog.c successfully without warnings, even when I give it the "-Wall -pedantic" options. It inserts a lot of extra spaces into the preprocessed output which is why I had to add the "ed" step to the build file so that my macro #defines would work. The compile step gives the usual messages with "-Wall -pedantic". The link step generates a warning about a name conflict between my BC() and a BC() from curses in the main system library. BUGS, I mean FEATURES: Arguments matching /[^2]22$/ cause the program to segfault after rapidly exhausting the available stack space. I could probably fix this by adding some special cases to the code, but [Blaine the Mono] it's quite a bit more exciting this way, don't you think?[/Blaine] Since the trailing "22" will be its own atom anyway, you can just insert a space in front of it if you run into this feature (e.g. "1 22" instead of "122"). Note: you may want to be sure you have a limit on stack size before trying this. I ran it on Debian GNU/Linux and discovered that I did not have a limit on stack size. Fortunately the OS killed the process after a while. This is consistent with the output of 1996/schweikh3 which reports "malloc lied! Killed by SIGKILL". Arguments matching /(?:.){10,}/ will produce visually incorrect results (e.g., generation 1 starting from 1111111111 will print as ": 1"), but there's nothing interesting about these arguments that can't be modeled using shorter runs of the same symbol. Arguments matching /(?:.){257,}/ may produce mathematically incorrect results (e.g., generation 1 starting from a string of 257 1's will be calculated as "11"); the remark from the previous paragraph applies here also. The program has exactly 2004 IOCCC countable bytes. It's tailor-made for this competition! Although there are no comments, the program still manages to be self- documenting in an obscure fashion (periodic table of the (non-primordial) elements, "OBFUSCATED C"). The program is obfuscated: By invoking cpp twice By the use of #defines By the use of typedefs By the use of global constants By the use of global functions that replace the standard ones (I particularly like substituting At (Astatine, highly unstable) for abort.) By overloading symbols (For example, B is a type, a structure, a structure member, a macro parameter, and a variable.) By standard tricks like writing '\0' as 1["3"] or 2["22"] By using ternary operators to avoid "else" By using the Sb function to create subbuffers from either end By naming everything after chemical elements or pairs of elements Because it says so in the source. ;)