This is an application that will generate tickets to play bingo 90(housie bingo, tombola).
#Hot to run ./gradlew resolveAndLockAll --write-locks to create gradle.lockfile
./gradlew clean build to compile
./gradlew run to run it!
Generate a strip of 6 tickets are created as strips of 6, because this allows every number from 1 to 90 to appear across all 6 tickets. If they buy a full strip of six it means that players are guaranteed to mark off a number every time a number is called. A bingo ticket consists of 9 columns and 3 rows. Each ticket row contains five numbers and four blank spaces Each ticket column consists of one, two or three numbers and never three blanks. The first column contains numbers from 1 to 9 (only nine), The second column numbers from 10 to 19 (ten), the third, 20 to 29 and so on up until The last column, which contains numbers from 80 to 90 (eleven). Numbers in the ticket columns are ordered from top to bottom (ASC). There can be no duplicate numbers between 1 and 90 in the strip (since you generate 6 tickets with 15 numbers each) Please make sure you add unit tests to verify the above conditions and an output to view the strips generated (command line is ok).
Try to also think about the performance aspects of your solution. How long does it take to generate 10k strips? The recommended time is less than 1s (with a lightweight random implementation)
The execution time unfortunately is bigger than 1 sec, however less than 4 secs.
I am using a ThreadLocalRandom which should perform better while multithreading.
Preferred an ExecutorService which should be good to run independent tasks in parallel. I have used a WorkStealingPool which seems performing slightly better than the FixedThreadPool,also tried to apply the formula Number of threads = Number of Available Cores * (1 + Wait time / Service time) which considering 1 ms waiting and 250 ms service time won't go far from 1,004 so preferred to just keep the number of threads.
Only part that is using a parallel stream is while creating tickets from columns.
For Row and Column internal implementation is using an array, trying to improve performances.
Jprofiler could only highlight issue related with logback which in the end is not making execution time that bad.
Maybe uses of stream could be replaced with for loops and operation on arrays, however that will make the code very hard to maintain and for me this was a good exercise for practicing functional programming .