The overall goal of the project is to add support for storing histogram statistics on disk in JSON format. As described in the related Jira page, storing histograms in comes with some benefits like:
- Allow DBAs to examine (and modify) the histograms. For instance, they can take advantage of existing JSON_* functions like JSON_LENGTH and JSON_EXTRACT to examine the generated histograms.
- Improve the histogram precision especially for VARCHAR data types.
- Provide a base for other histogram formats such as most-common-values (MCV).
NB: Below, "binary histograms" is used to refer to both single and double precision histograms (i.e SINGLE_PREC_HB and DOUBLE_PREC_HB).
- MDEV-21130 - Tracking JIRA page for this project's issue.
- MDEV-26125 - Sample bug demonstrating a limitation in the current binary histogram.