How to use Ruby DBI's 'select_all' vs 'execute-fetch/each-finish'

Question

Here is my sample code for using DBI:

dbh = DBI.connect(\"DBI:Mysql:host=#{server};database=mysql\", user, pass)
rows = dbh.select_all(\"SHOW TABLES\")


        

Answer 1

I guess that dbh.select_all returns an instance of Enumerator, which yields the same row on each iteration. See this pseudocode to understand what I mean:

def select_all(query)
  db_row = Row.new
  reader = @connection.execute_reader(query)
  Enumerator.new do |yielder|
    until reader.end?
      db_row.populate_from(reader)
      yielder.yield db_row
      reader.next!
    end
  end
end

Thus, if you're using select_all without block, an Enumerator will be returned, which basically yields the same db_row object.

This is just a guess, but, I believe the truth is nearby.

---

The source code fetch method definition says that I was wrong, because @row is duplicated on each iteration. Well, probably the error is somewhere up the stack.

Answer 2

The same is also happenning when I query an MS SQL DB using DBI.connect("DBI:ODBC:Driver={SQL Server};...")

My work around was to forcably convert the DBI::Row to an array:

sth = dbh.execute "..."
begin
  return sth.map{ |row| row.to_a }
ensure
  sth.finish
end

I'll bet $1,000 that the issue is to do with buffer reuse - a possible 'performance' enhancement that has had underirable side effects!