"mississippi" gives "issi" instead of "ssi" as longest repeated substring

[graste]

Expected ssi instead of issi for input mississippi.

[shrink0r]

Isn't "issi" actually the longest repeated substring?
It is in there twice, as is "iss", but because "issi" is longer it wins.

[graste]

I'd say "ississi" doesn't contain "issi" twice.

[shrink0r]

The way it is implemented at the moment, just gives the longest repeated substring path of thedeepest internal node. This may lead to an overlapping result.
Can be seen here for the "mississippi" example:
http://www.allisons.org/ll/AlgDS/Tree/Suffix/ (see the section "Longest Repeated Substring")
Not being able to determine the non-overlapping repeated-substring can be considered a bug, because the method name is kinda ambigious concerning the result.
@graste Do you think either a second method or an additional param '$overlap = false' make sense (1aa8b93)?

[graste]

My expectation obviously was "no overlapping" as there are not "enough" characters to actually write down the repeated substrings. That's why I'd prefer a default of `allow_overlapping = false``. As this is a library l'd prefer two methods (and maybe a private one with the boolean flag to not repeat yourself). Not sure about naming though. :-D

[MrHash]

Wouldn't 'iss' be the longest repeated substring matched ahead of 'ssi' in a non-overlapping context?

[shrink0r]

@MrHash: yes, as the substrings are built from prefixes of each suffix (l2r), "iss" will win over "ssi".