datasketches-cpp icon indicating copy to clipboard operation
datasketches-cpp copied to clipboard

Request: Add Binary Layout Diagram for HLL Sketches (HLL4 / HLL6 / HLL8)

Open SiddheshDhinge opened this issue 1 month ago • 0 comments

Hi DataSketches team,

I’m working with the HLL sketch family (HLL4, HLL6, HLL8) and need a clear understanding of the serialized binary layout, including:

  • header structure
  • Register arrays
  • AuxHashMap structure for overflow
  • Alignment rules and padding (if any)

It would be extremely helpful to have:

  1. A visual diagram or table showing offsets, byte/bit boundaries, and field descriptions for each HLL type.
  2. Confirmation that the Java and C++ implementations share an identical on-wire layout.
  3. A link to any existing internal documentation if it already exists.

This would make it much easier for developers working on:

  • Custom storage systems
  • Memory compaction
  • Cross-language interoperability

If such a document exists, could you point me to it? If not, would the team consider adding a binary layout diagram to the HLL section of the documentation?

Thanks

SiddheshDhinge avatar Nov 23 '25 10:11 SiddheshDhinge