Dedupe using length doesn't appear to be working
To reproduce on dmaster
- create a rule that checks email, postal code, or street address on first X characters
- run dedupe rule and see random (or perhaps name dedupe) certainly not by email
This is how 'length' feature in dedupe is documented pretty clear it is intended to work this way, and IIRC worked as recently as two years ago @eileen please tag as regression if appropriate