When importing a CSV file, the presence of the text delimiter should cause the 
cell to be automatically treated as text.  It is, after all, the primary purpose 
of a text delimiter to indicate that the value is text.

While it is true that you can specify that a column is text during the import 
process, this can be a real pain and error prone.  I frequently want to keep the 
file in CSV format so it can be shared with other applications -- so I may need 
to open the same CSV file in OOo many times.  Having OOo default to the correct 
handling would make this much easier.

A good example of why this important is handling of postal codes, especially US 
ZIP codes.  These codes often start with 0, and if formated as a number any 
leading 0's get dropped.  Another example is years formated as 2 digits (e.g. 
'07' for '2007') -- again, if treated as a number, the leading 0 is dropped.

It is easy to convert a column from text formating to numeric formating -- but 
it is relatively difficult to convert a numeric column back to text once the 
leading 0's have been lost. 

An alternative solution would be to treat any value with a leading 0 as text.  
This is less elegant, but might be more compatible with a broader range of CSV 
exporters.

Either of these solutions probably also demands that there be an option in the 
import to explicitly format a column as numeric.  In other words, if the default behavior switches to text under any circumstances, then there needs to be a way 
to explicitly override that default.