Columns are difficult to work with from Python

[rbasralian]

It is tough to work with column data/column types from Python:

1. There's no way to get a list or set of column names (e.g. t.getDefinition().getColumnNames() in Java)
2. There's no way to get a Map of columns to column definitions or even data types (e.g. t.getDefinition().getColumnNameMap() in Java)
3. The only way of extracting data from a column are too roundabout — we should have a method for pulling data out of columns, or just support indexing the table like myTable['column_123'] (like t.getColumn("column_123").getDirect() in Java).

We should have Python analogues for each of these — all of these cases come up frequently.

Examples for each of the above:

1. Iterating over column names to do something, or doing something if a specific column is missing/present
2. Converting a column if its data type is not what you want (e.g. after reading from files)
3. Iterating over values in a column

[chipkent]

I agree with (1) and (2).

I do not agree on (3). Great effort has gone into providing access that is properly locked and consistent. This can currently be done using to_numpy(t, ["X"]). This is very direct and straightforward.

[chipkent]

Also, there is a plan for a more efficient to_numpy(), but it apparently does not yet have a ticket. It is item 54 in the python master plan.