#3428 replaced the parallel map2 implementation with its "old" version (based on both; because it's faster). However, that PR also contains another custom implementation (2c4bc68), which seems like it could be faster than both, but apparently it is not. But, if traverse is optimized in cats, it should be retested. This issue is so that this is not forgotten.
#3428 replaced the parallel
map2implementation with its "old" version (based onboth; because it's faster). However, that PR also contains another custom implementation (2c4bc68), which seems like it could be faster thanboth, but apparently it is not. But, iftraverseis optimized in cats, it should be retested. This issue is so that this is not forgotten.