Hi, I am experimenting a migration of an M$ Access database to postgresql in a Debian GNU/Linux system. Based on some reading, I've approached it as follows: 1. Using mdbtools, export the tables from Access to *.csv files, sanitizing file names and column names. 2. Using unixodbc and GNU R's package RODBC, import the tables to an empty database in postgresql; i.e. create a database in postgresql, read the tables into R for cleaning up mbdbtools' job and let RODBC import the tables into postgresql, specifying the data type for date/time/timestamp columns. Checking the results in pgadmin, this proceeded fine, but now that I want to specify the primary and foreign keys in the tables, I see that the columns needed for this were imported as int4 data type. I would like these to be automatically sequenced, so need them to be 'serial'. Going into "properties" for these columns in pgadmin, the 'serial' option is not available in the "data type" pull-down menu. Is this not possible? Can the 'serial' data type be specified during import. Thanks in advance. Cheers, -- Seb
