Open mizvyt opened 4 years ago
this, #38, #31 might all be linked. I'll take a look.
I was able to reproduce this finally, and noticed this stays even after #40. (Using bf.approx_len gets us correct length...) but need to dig deeper.
I reckon the problem might be around self._bf.count_correct
and self._bf.elem_count
in the cython class.
Which begs the question: why not use approx_len
all the time?
For reference, this is the patch:
https://github.com/axiak/pybloomfiltermmap/pull/47/files#diff-a16b2c8dba66561e6fdef61b6efe2b04
Can you not merge that patch into your repository and that publish the updated version to the pypi?
Issue in detail here: https://github.com/axiak/pybloomfiltermmap/issues/31
A fix is referenced, but I'd like to still think over the options.