Hi @Nabil AMEZIANE​ ,

What you were doing works actually...

You just need to fetch the latest version into a variable like below

x= spark.sql("""select max(version) from (desc HISTORY <table-name>)""").collect()[0][0]

Be sure to check the operation type too in the history table if there are DELETE operations.

Now you have a integer x with latest version number available. For previous version all you need to do is to do a x-1. Then you have both version numbers.

Now pass these variables into your sql query reading those versions. I would do it as below

spark.sql("""
select * from table @v{0}
minus
select * from table @v{1}
""".format(x, x-1))

Hope this helps...Cheers