Codecov Report

Merging #103 (0fcac2b) into master (ccb39ed) will increase coverage by 0.01%.
The diff coverage is 100.00%.

@@             Coverage Diff              @@
##             master     #103      +/-   ##
============================================
+ Coverage     56.33%   56.34%   +0.01%     
  Complexity     1176     1176              
============================================
  Files           149      149              
  Lines          7992     7987       -5     
  Branches        766      766              
============================================
- Hits           4502     4500       -2     
+ Misses         3246     3242       -4     
- Partials        244      245       +1     

Help us with your feedback. Take ten seconds to tell us how you rate us.

Could we have a simple benchmark test?

Could we have a simple benchmark test?

Performance is not the main concern here.
Exception-free is more important.

If you find this change is impacting performance later,
you can change the method body to

try {
    return deserializeBitMap(serializeBitMap(bitmap));
} catch(IOException e) {
    // not expected
}

Just look at the body of serializeBitMap and deserializeBitMap,
it feels they are more costly and error-prone.

And explicitly throwing RuntimeException is not a good practice,
because they are difficult to catch (without catching real RuntimeException).

Could we have a simple benchmark test?

Performance is not the main concern here. Exception-free is more important.

If you find this change is impacting performance later, you can change the method body to

try {
    return deserializeBitMap(serializeBitMap(bitmap));
} catch(IOException e) {
    // not expected
}

Just look at the body of serializeBitMap and deserializeBitMap, it feels they are more costly and error-prone.

And explicitly throwing RuntimeException is not a good practice, because they are difficult to catch (without catching real RuntimeException).

Sadly, we care the performance here. We need performance test to ensure that the pr won't bring worse performance. Bitmap method or will do many operations, it may cost much time, I can't ensure the performance easily.

It turns out the new method is 15x faster than the old one.
I'll include the following test in this PR.

Great work!

Could you use some random number to test them again? Because bitmap will compress the ordered number, the size of bitmap will be smaller, we test in more situations, the result will be more persuasive.

Random ints

Random longs

I'll include the following test in this PR.

Removed this test because it might be flaky due to resource sharing in the CI environment.

Thanks @jerqi for the review.