Character.compare('1','2'); //out: -1
This output means that that "2:..." will go before "1:..." as required in the solution;

so clever done,i was stuck by this kata

Love it

Guys, correct me if im wrong but this solution proposes to use regex-based replaceAll() in a loop, which compiles and applies regex for every single char, which is, well, slow.

If we have "1:bbb" and "2:aaa" then result will be "1:bbb/2:aaa", when the right answer is "2:aaa/1:bbb"

It is compared from the beginning, including the 1,2 and =,
if it is the same,then ":",same,then the letter lexicographic order

This is very clever and compact but certainly not something you would want to put into production unless it is called very rarely. Using regular expressions comes at a cost. Clever: Yes! :-) Best practice: No.

Awesome solution!