Re: Sudoku solver

The die is used for control flow here -- there's a deep recursion going on, and once a solution is found, all of these recursive subs can be terminated at once. That's what die does.

And warn doesn't -- it just prints a warning, and doesn't abort the current routines. So it makes the solver search for more solutions even after a solution is found. Which is why it takes much longer.

A possible (still hacky) approach is to do the BLOCK form of eval to catch the exception, print the result and then start the recursion anew with the next Sudoku.

Perl 6 - the future is here, just unevenly distributed

Comment on Re: Sudoku solver Download Code

Replies are listed 'Best First'.
Re^2: Sudoku solver by davido (Cardinal) on Aug 11, 2012 at 15:49 UTC
I actually found a non-obfuscated, non-golfed Sudoku solver in C++ once that used a similar technique; recurse, and then throw an exception for any branch that didn't pan out. There were exceptions at multiple levels, each being caught to wipe out branches. Worked pretty well, though I could faintly hear Dijkstra moaning when I ran it. I wish I could find and link to it now. It was interesting. Dave	[reply]
Re^2: Sudoku solver by Anonymous Monk on Aug 11, 2012 at 11:33 UTC
maybe like `print forgetR( ... ); sub forgetR { local @A; if( 1 == @_ ){ @A = split //, @_; } else { @A = @_; } eval { R(); }; warn "CAUGHT DEATH $@ "; return @A; }` [download] forgetR takes either a line (like from __DATA__) or a list	[reply] [d/l]