-
Type:
Engineering Test
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: None
-
None
-
Storage Execution
-
Fully Compatible
-
Storage Execution 2026-10-12, Storage Execution 2026-11-09, Storage Execution 2026-11-23
-
200
-
None
-
None
-
None
-
None
-
None
-
None
-
None
An idea that has come up repeatedly is to create a corpus of edge case BSON documents that we can use for deterministically testing the database. The linked BF is an example where that could have helped, as empty array vs sub-document is a known cause of subtle issues. Other examples would be very large documents, very large keys, very many keys, different geo data formats, very deeply nested documents, strings with non-UTF8 contents. We would want all BSON types, including deprecated (but allowed) ones. We would want all "special" values: -0.0, infinities, NaNs, subnormal numbers, decimal128 edge cases for all these as well as trailing zeroes. Integers that cannot be represented exactly in floating point. Empty field names. Very long field names.
For indexes we would want to try and build all different kinds, sparse, unique, hashed, multi-key, ascending and descending. Compound indexes over these, including the maximum number of fields. We would want edge case BSON in the _id field.