Prior to version 4.1 smart names were calculated for a subset of records when you do an export.
A problem that has been brought to my attention is duplicate name resolution. GSAK currently calculates a smart name and if already exists changes the name to ensure it is unique.
The problem here is that this duplicate name resolution is at the subset level and not the database level. For example, if you set a filter on finds and do some export, then do a filter on not found and do another export, it is now possible that some of the smart names could be the same across these exports. This can cause problems if you are using smart names when you send waypoints down to something like cachemate, because you usually split your exports up into several categories before exporting.
Solution:
Before doing the export GSAK now calculates all smart names for the whole database. This will add a bit of extra time up front to all exports - perhaps only noticeable when you have a large database but only export a small subset. This change means that calculated smart names will always be the same for the current snapshot of the database.