Currently if we have a testcase of size 4246 and our first reduction pass of size 4096 is successful we then have a testcase of size 150. Then we will try to remove the remaining part and then reduce the chunk size by 1/2. The problem here is we will then try with 2048, 1024, 512, 256, before getting to 128. This means we will test with an empty testcase 5 times. This can be optimized. PR to come.
Here is an example log of the issue:
Intermediate files will be stored in tmp129/.
The original testcase has 4246 lines.
Checking that the original testcase is 'interesting'...
Interesting: Assertion failure: blah... blah... blah...
Updating poll timeout to: 10
Updating max timeout to: 10
Interesting: Assertion failure: blah... blah... blah...
Removing a chunk of size 4096 starting at 0 of 4246 was a successful reduction :)
Uninteresting: no failure detected
Removing a chunk of size 4096 starting at 0 of 150 made the file 'uninteresting'.
Halving chunk size to 2048
Uninteresting: no failure detected
Removing a chunk of size 2048 starting at 0 of 150 made the file 'uninteresting'.
Halving chunk size to 1024
Uninteresting: no failure detected
Removing a chunk of size 1024 starting at 0 of 150 made the file 'uninteresting'.
Halving chunk size to 512
Uninteresting: no failure detected
Removing a chunk of size 512 starting at 0 of 150 made the file 'uninteresting'.
Halving chunk size to 256
Uninteresting: no failure detected
Removing a chunk of size 256 starting at 0 of 150 made the file 'uninteresting'.
Halving chunk size to 128
Uninteresting: no failure detected
Removing a chunk of size 128 starting at 0 of 150 made the file 'uninteresting'.
Interesting: Assertion failure: blah... blah... blah...
Removing a chunk of size 128 starting at 128 of 150 was a successful reduction :)
Currently if we have a testcase of size 4246 and our first reduction pass of size 4096 is successful we then have a testcase of size 150. Then we will try to remove the remaining part and then reduce the chunk size by 1/2. The problem here is we will then try with 2048, 1024, 512, 256, before getting to 128. This means we will test with an empty testcase 5 times. This can be optimized. PR to come.
Here is an example log of the issue: