Re: [SPOILER] Perl Quiz of the Week #24 (Turing Machine simulation)

Quoting "Michael C. Toren" <[email protected]>:

> While using your implementation to debug Turing programs, I discovered an
> additional small bug.  If in a given state there is no state table entry
> matching the tape cell under the head, the program exits as expected, but
> only after erasing the cell (setting it to undef).  For example, if the
> Turing program:
>
> 	A 1 B 1 L
>
> is given an initial tape of "0", it will output "" rather than "0".  One
> possible workaround is:

Note that it gets even worse if there's trailing underscores - this program,
with a blank initial tape, shows a problem as well:

a _ b 1 R
b _ c _ R
c _ d _ R
d _ e _ R
e X stop X R

That should print "1", but mjd's original prints "1___".  And I 
personally think
it's easier, and the result is more readable, to make this change:

24c24
< while (my $ttab = $transition{$STATE}) {
---
> while (my $ttab = $transition{$STATE}{$TAPE[$HEAD]}) {
37c37
<   ($STATE, $TAPE[$HEAD], $inc) = @{$ttab->{$TAPE[$HEAD]}};
---
>   ($STATE, $TAPE[$HEAD], $inc) = @{$ttab};

But yes, the debugging mode is nice - without it, I would have given up on the
parens program when my first attempt failed.  (Though I only had 5 bugs to
track down, and 4 of those were transcription errors from the stack of paper
where I'd worked it out to begin with.  This is more a reflection of the fact
that regexp substitutions often translate easily to Turing machines than
anything else)

--
@/=map{[/./g]}qw/.h_nJ Xapou cets krht ele_ r_ra/;
map{y/X_/\n /;print}map{pop@$_}@/for@/