Daniel Lemire

Implement the AVX-512 versions of the following functions

13

Starting with version 1.0.0 of the library we have AVX-512 routines, but we are still missing many optimizations. The following functions related to array containers may be upgraded to versions...

Support serialization operations on big-endian platforms in little-endian format

1

Big-endian platforms are effectively gone. Nevertheless, as a matter of principle, we would like to support them for the IBM mainframe users out there. Currently, if you are on a...

Implement roaring_bitmap_internal_validate

4

When deserializing a bitmap, it is possible that the result might be invalid. This could happen because there was data corruption. The deserialization could still generate a bitmap without failure,...

Add getContainerCount() method

4

**Is your feature request related to a problem? Please describe.** In some instances, we want to sort bitmaps with respect to the number of containers they hold. See https://github.com/RoaringBitmap/RoaringBitmap/issues/608#issuecomment-1364722016 **Describe...

Instead of failing with "Too many iterations. Are all your keys unique?", I recommend filling in the fingerprint buffer with 1s. Example in Go: ```Go if iterations > MaxIterations {...

Fixes https://github.com/glitzflitz/pyxorfilter/issues/10

Integrate in Valkey

2

This should be proposed to https://github.com/valkey-io/valkey as it is obvious that it has generated sustained interest for many years.

Load without copy?

1

Current code makes a copy of the data upon import. In the case where the data comes from disk, this would render memory-file mapping useless.

The ``checkflush()`` test does not pass with WAH (see ``tests/unit.cpp``) which means that arbitrary WAH inputs are not supported.

Leo's populate_hits_avx

1

I took Leo's PR and reverted back the change in cache size, to isolate the effect of his new populate_hits_avx. (It is hard to reason about multiple changes at once.)...