Found bugs in Elf and Elf+

[urbanideskonstantin]

I tested both, Elf and Elf+ on my large time series data set (+100GB) and I got exceptions when compressing some of the floating-point values. In one case, the compressed floating-point value could not even be decompressed losslessly anymore.

Here are some of the values that lead to the bugs:

Elf

• 1.530857361106962E18

Elf+

• 1.530857361106962E18
• 3.679935659961374E17

I therefore provided a fix for both algorithms on my local branches.

[IrisLoveLeo]

Thank you for pointing it out. I will ask my teamates to check it.