InputStreamReader isr =new InputStreamReader(str, encoding);
ArrayList<String> al =new ArrayList<String>(128900);
char[] ar =newchar[26];
while(isr.read(ar) != -1){
al.add(new String(ar);
}
Why is that in my case of a 2.8 MB text file, when just doing a System.currentmillis I can faster times for a char array of 24-26 ?
What is the correlation?
Here is the full code
It reads a txt file with 128900 words in UTF-8
InputStreamReader isr = new InputStreamReader(str, encoding);
//__log.debug(isr.markSupported());
//128900
ArrayList<String> al = new ArrayList<String>(128900);
char[] ar = new char[24];
char[] word = new char[26];
int offset = 0;
int length = 0;
while( (length=isr.read(ar)) != -1) {
for (int i = 0; i < length; i++) {
if(ar[i] == '\r' || ar[i] == '\n') {
if(offset != 0) {
al.add(new String(word,0,offset));
offset = 0;
}
} else {
word[offset] = ar[i];
offset++;
}
}
}
This code executes in 600ms with a value of 24-26 in the array ar
If you change that value to more or less, code executes more slowly
NOTE: we know advance that the max length of a word is 26 char that why we have a word array of 26
Using a [url=http://java.sun.com/j2se/1.5.0/docs/api/java/io/BufferedReader.html]BufferedReader[/url] might be a good idea, IMHO.
A buffered reader provides a useful readLine method.
Anyway, regarding the impact of buffer size on performance, I guess you'ld better choose a power of two (e.g. try comparing using 1000 and 1024.)
However, I'm not sure you'll reach significantly better performance with your approach, comparing to using a BufferedReader.
And your code would certainely look simpler and clearer.
