That's a negative on cache.py. It's from a nightly build pulled about 5 days ago. The csv in question is about 150k records. (have pruned it down to a test of 10 records, and they import OK). Has about 140 fields, quite a few nulls, and have compensated for the adapter's postgres decimal precision issue. (its the largest table in the app, and others appear to import OK)
Still building a u12.10 environment under vmware to test (linux skills are rusty but coming back) --
