Shortest program whose concatenated program+output is a permutation of the 95 printable ASCII characters

Question

Write a program that's as short as possible (as measured by byte-count), takes no input, contains no comments, and outputs a character string/sequence, such that concatenated program+output is exactly 95 characters long and contains every printable ASCII character exactly once, i.e. program+output is a permutation of the 95 characters with ASCII codes 32-126:

 !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~ 

In the event of a tie, the winner is the entry whose (95-character) concatenated program+output is closest to the above string, as measured by the Levenshtein edit-distance (i.e. the number of single-character deletions, insertions, or substitutions required to put it into ASCII order).

Answer 1

GolfScript, 14 characters

{`),32>46-^}.~

Output:

 !"#$%&'(*+/015789:;<=?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]_abcdefghijklmnopqrstuvwxyz|

Answer 2

Perl, 89 characters

This is the best I've been able to do so far:

q< !"#%&'()+/8:?@ABCDEFGHIJKLMNOPQRSTUVWXYZ\^_`abcdefghjklmosuvwxz|>;
print~-$=.73*2,y[]{}

For those not conversant in Perl minutae, the two arguments to the print statement could be rendered as ~(-($=)) . (73 * 2) and tr///.

The program outputs 6 characters:

591460

Unfortunately every single iteration construct that I can think of in Perl (for, while, until, map, grep) shares at least one character with print. If newlines were allowed to appear in the output, then I could use Perl 5.010 to eke out an 87-character solution:

q{ !"#%&'*,-./0:;>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^`bcdfgjkmnoprtuvxz|~}while($_+=say)<9

Answer 3

PHP 67 bytes

<?=U4eB1gkFdA6J9snZD2IE8y5PhwQV^xHGmqYMfNjKpbR3vcLalCWi0_TtrOSXouz;

Output:

-|"/@>& *+}:[!],'~(){.\`7#%$

It seems like there should be a 65 byte solution, outputting 30 chars instead of only 28, but I couldn't get it to line up. The 6 unused characters OSXouz have been tacked onto one of the string literals instead.

Edit: Upon further reflection, it is not possible to generate more than 28 characters in this manner. UPPER ^ lower will always result in a character between 32 and 63, so all 10 digits are required for @[\]_`{|}~. Four of the remaining 22 are used in the script itself, leaving only 18 which can be obtained with letters alone. The tiebreaker could be improved significantly, though.

Answer 4

Ruby, 91 chars

%{ !"#$&\',-/450:<=>ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_`abcefghijklmnqvwxyz|~};puts(?@‌​.ord*98+1)

Outputs 6273.

%{...} trick stolen from Jan Dvorak

Answer 5

Perl, 61 characters

print$:x8^CJIBQTOWAEULHVGFNMKRPZY,q<#"/;*g~?|={.>&+u15m97ws26

Output:

c@db[yo]le_ah\jfD`kX}zS-! %)('v304

$: defaults to " \n-". x8 repeats it eight times; the 24-character result is then bitwise-xored with a 23-character barestring (the last - is left as-is). Second part is a 12-character q-string bitwise-and a 10-character barestring (the last 2 characters are discarded).

Answer 6

Ruby, 95 characters, 6 edits

%{ !"#$&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz|~}

Yep. That's it. Just a free-floating string literal that is never printed. Ruby supports string literals (borrowed from Perl, I think) in the percent-type-delimiter form - the delimiter is any non-alphanumeric character, and if the opening delimiter is an opening bracket (any of [({<), the closing delimiter is the corresponding closing bracket (nested brackets are allowed).

I realise this is more of a baseline submission, but unless someone manages to actually print something or unless there's a language with a two-char freeform string literal with a different opening and closing delimiter, this is the best there is.

Answer 7

Tcl 96, edit distance 12

if 0    {!"#$%&'()*+,-./123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdeghjklmnopqrstuvwxyz|~}