Just for fun, I decided to map out a bit of the region beyond pack's normal range. The results were... interesting.
| ord |
pack |
len |
chr |
| 0 | 64 | | |
| 112 | 65 | | p |
| 97 | 66 | 89 | a |
| 110 | 67 | | n |
| 105 | 68 | | i |
| 99 | 69 | 93 | c |
| 58 | 70 | | : |
| 32 | 71 | | |
| 109 | 72 | 97 | m |
| 101 | 73 | | e |
| 109 | 74 | | m |
| 111 | 75 | 101 | o |
| 114 | 76 | | r |
| 121 | 77 | | y |
| 32 | 78 | 105 | |
| 119 | 79 | | w |
| 114 | 80 | | r |
| 97 | 81 | 109 | a |
| 11 | 82 | | p |
| 0 | 83 | | |
| 0 | 84 | 113 | |
| 0 | 85 | | |
| 0 | 86 | | |
| 0 | 87 | 117 | |
| 0 | 88 | | |
| 0 | 89 | | |
| 0 | 90 | 121 | |
| 0 | 91 | | |
| 0 | 92 | | |
| 0 | 93 | 125 | |
| 0 | 94 | | |
| 0 | 95 | | |
| 67 | 96 | 129 | C |
| 97 | 97 | | a |
| 110 | 98 | | n |
| 39 | 99 | 133 | ' |
| 116 | 100 | | t |
| 32 | 101 | | |
| 108 | 102 | 137 | l |
| 111 | 103 | | o |
| 99 | 104 | | c |
| 97 | 105 | 141 | a |
| 108 | 106 | | l |
| 105 | 107 | | i |
| 122 | 108 | 145 | z |
| 101 | 109 | | e |
| 32 | 110 | | |
| 116 | 111 | 149 | t |
| 104 | 112 | | h |
| 114 | 113 | | r |
| 111 | 114 | 153 | o |
| 117 | 115 | | u |
| 103 | 116 | | g |
| 104 | 117 | 157 | h |
| 32 | 118 | | |
| 97 | 119 | | a |
| 32 | 120 | 161 | |
| 114 | 121 | | r |
| 101 | 122 | | e |
| 102 | 123 | 165 | f |
| 101 | 124 | | e |
| 114 | 125 | | r |
| 101 | 126 | 169 | e |
| 110 | 127 | | n |
| 99 | 128 | | c |
| 101 | 129 | 173 | e |
| 0 | 130 | | |
| 0 | 131 | | |
| 0 | 132 | 177 | |
Besides the panic: memory wrap and Can't localize through a reference, one other thing stands out: pack u78 uses a space as the delimeter. In cases where you can't quite get your code to fit into a u45 format (starting with an m), it's a viable alternative, and has actually been the shortest construction I've been able to find for more than one challenge.
The run of chr(0) from u83 - u95 is unfortunate, though. An eval'd string beginning with the null character seems to terminate immediately, without error, for whatever reason.