It would be great to be able to make requests through thrift for a subset of a row via a start/stop column pattern. Below are three different read patterns any of which would significantly help reading columns through thrift.
From a single request perspective it would be nice to be able to specify a start/stop col to read in one call everything for that row with the specified start/stop column.
getRowWithColumnRange(tableName, row, startColumn, stopColumn)
It would also be very useful to set up a column scanner where we could stop at any point. Basically instead of scanning rows we would scan columns for a given row. In the below example we would set up a scanner to scan the columns and we could read xx cols back at a time and stop any time we had "enough" data.
colScanID = colScannerOpenWithStop(tableName, row, startColumn, stopColumn)
Lastly is would also be useful/flexible to expose columns to the current row scanner process to basically be able to cross rows and still only receive XX column values. The new scannerGetCols method below would return 1000 col values at a time instead of XX rows at a time like a normal scannerGetList. This scanner would also be able to cross rows seamlessly if needed.
scanID = scannerOpen(tableName, row)