3 ms·

I am not sure this really is "amb" in all its glory. The incredible power of amb comes from the fact that some of the computations involved may be crashing or d

• colanderman 7y ago
  C/POSIX permits a simple (albeit wildly inefficient) implementation that handles crashing: #include <stdlib.h> #include <unistd.h> #include <sys/types.h> #include <sys/wait.h> // Return an integer in the range [0, n) which // causes execution to succeed. unsigned amb(const unsigned n) { if (n == 0) _exit(42); for (unsigned i = 0; i < n - 1; ++i) { const pid_t child = fork(); if (child < 0) abort(); if (child == 0) return i; int wstatus; if (waitpid(child, &wstatus, 0) < 0) abort(); if (WIFEXITED(wstatus) && WEXITSTATUS(wstatus) == 0) _exit(0); } return n - 1; } You could of course parallelize this to handle nontermination, at the expensive of potentially many concurrent processes for a high n. You can make other fun time-traveling functions following the above pattern. My favorite I call the "Groundhog Day" function, or "Cause and Effect" for Star Trek fans: when first called, it returns zero; if the program later fails, it is restarted from that function call, which returns the exit code. (I'm not really sure how useful this is.)
• giornogiovanna 7y ago
  How does amb "in all its glory" handle an endless loop? Any solution I can think of seems prohibitively expensive.