xget seems to be a great idea, but for me in pycassa 1.6.0, it is failing every time after iterating through 1024 values with 'an unexpected error while tokenizing input'. The traceback goes to a TypeError: unsupported operand types for &: 'int' and 'str' in line 324 of marshal.
A similar paging version of insert would be great too since inserts typically timeout when writing large dictionaries to a single row (500k+ columns).
Fix xget paging for non-string comparators
column_start and column_finish were being packed twice,
which broke when the comparator had a non-idempotent pack
Thanks! Looks like the problem was for non-string comparator types only.
Regarding insertions, batch module can handle this fairly well, but I've opened #140 to try to make it more flexible for inserting wide rows.
You're right, it does seem to be working fine for string comparators. As for #140 - thanks!