static int longest_length = -1;

No. Static is useless here and makes you method much less usable. A non-static variable would be better, but actually, why not return a result???

Better, return the substring, the length can be easily obtained from it.

Separated printing and computation, good.

if(index >= str.length()) return;

Spacing (after if). Also according to conventions, you should always use braces (I personally don't).

LinkedHashMap<Character,Integer> map = new LinkedHashMap<Character,Integer>();

Use Java 7 diamonds or Guava's Maps.newLinkedHashMap() to save yourself repeated generics.

You need no linked map.

calc(str,last_index+1);

I find the recursion confusing here. You really do nothing, but a simple loop, so use a loop.

The algorithm indeed runs in O(n*n), which can be shown with a string like

abcde.... abcde....

where you always run through half of the string. I was assuming an unlimited alphabet here, more precisely the complexity is O(n*alphabetSize) as no runs can be longer than alphabetSize.

A much better O(n) algorithm would keep track of the starting and ending positions. Wherever you see a duplicate, advance the start. Sounds damn simple.

static int longest_length = -1;

No. Static is useless here and makes you method much less usable. A non-static variable would be better, but actually, why not return a result???

There's no longest_length in Java. Maybe longestLength.

Better, return the substring, the length can be easily obtained from it.

Separated printing and computation, good.

if(index >= str.length()) return;

Spacing (after if). Also according to conventions, you should always use braces (I personally don't).

LinkedHashMap<Character,Integer> map = new LinkedHashMap<Character,Integer>();

Use Java 7 diamonds or Guava's Maps.newLinkedHashMap() to save yourself repeated generics.

You need no linked map.

calc(str,last_index+1);

I find the recursion confusing here. You really do nothing, but a simple loop, so use a loop (Even a goto would be clearer than recursion here).

The algorithm indeed runs in O(n*n), which can be shown with a string like

abcde.... abcde....

where you always run through half of the string. I was assuming an unlimited alphabet here, more precisely the complexity is O(n*alphabetSize) as no runs can be longer than alphabetSize.

A much better O(n) algorithm would keep track of the starting and ending positions. Wherever you see a duplicate, advance the start. Sounds damn simple.

static int longest_length = -1;

No. Static is useless here and makes you method much less usable. A non-static variable would be better, but actually, why not return a result???

Better, return the substring, the length can be easily obtained from it.

Separated printing and computation, good.

if(index >= str.length()) return;

Spacing (after if). Also according to conventions, you should always use braces (I personally don't).

LinkedHashMap<Character,Integer> map = new LinkedHashMap<Character,Integer>();

Use Java 7 diamonds or Guava's Maps.newLinkedHashMap() to save yourself repeated generics.

You need no linked map.

calc(str,last_index+1);

I find the recursion confusing here. You really do nothing, but a simple loop, so use a loop.

The algorithm indeed runs in O(n*n), which can be shown with a string like

abcde.... abcde....

where you always run through half of the string. I was assuming an unlimited alphabet here, more precisely the complexity is O(n*alphabetSize) as no runs can be longer than alphabetSize.

A much better O(n) algorithm would keep track of the starting and ending position. Wherever you see a duplicate, advance the start. Sound damn simple.

static int longest_length = -1;

No. Static is useless here and makes you method much less usable. A non-static variable would be better, but actually, why not return a result???

Better, return the substring, the length can be easily obtained from it.

Separated printing and computation, good.

if(index >= str.length()) return;

Spacing (after if). Also according to conventions, you should always use braces (I personally don't).

LinkedHashMap<Character,Integer> map = new LinkedHashMap<Character,Integer>();

Use Java 7 diamonds or Guava's Maps.newLinkedHashMap() to save yourself repeated generics.

You need no linked map.

calc(str,last_index+1);

I find the recursion confusing here. You really do nothing, but a simple loop, so use a loop.

The algorithm indeed runs in O(n*n), which can be shown with a string like

abcde.... abcde....

where you always run through half of the string. I was assuming an unlimited alphabet here, more precisely the complexity is O(n*alphabetSize) as no runs can be longer than alphabetSize.

A much better O(n) algorithm would keep track of the starting and ending positions. Wherever you see a duplicate, advance the start. Sounds damn simple.