Thanks for the explanation, I've got it. (Then consider my node as an explanation to ikegami's node.)
It's funny how many different ways people interpret similarity/resemblance, because that was exactly the reason why I chose to keep all the optional characters if the id already fit in the char limit. That way my code always keeps the under-limit ids identical (== more similar). Of course in other cases that's not the optimal choice.
I also thought of (but not implemented) a more generic way to decide which character to drop from the original id: provide the user a filter callback in which s?he can rate the characters (or substrings) considered, then drop the ones with the lowest rating (still from right to left). For example: [_ ] => 3, [A-Z] => 2, [a-z]=> 1, anything else => 0
And this is why I've collapsed the char-level suffix tree to the substring-level: to ease the access to substrings for the purpose of rating. And also because the structure of the tree in the substring-level form cannot interfere with the selection of (non-)ambiguous characters (as in choroba's remark above if I get it right).
Cheers
In reply to Re^3: Generate unique ids of maximum length
by rubasov
in thread Generate unique ids of maximum length
by lima1
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |