Implement PyHDK QueryExpr, QueryNode and HDK API except data import. #192
Conversation
Signed-off-by: ienkovich <ilya.enkovich@intel.com>
| @property | ||
| def size(self): | ||
| res = self.c_node.size() | ||
| # For scans we hide virtual column |
There was a problem hiding this comment.
I didn't follow closely, c_node.size() isn't this a call to C++ API? And if so, doesn't it already contain the logic for subtracting 1 in case of a scan?
There was a problem hiding this comment.
In C++ we actually do not hide the virtual column in such a manner. So we do all these modifications for virtual columns in Python.
There was a problem hiding this comment.
Do you think this should be the user's responsibility? I'd think it's a rather specific feature for modin, isn't it?
There was a problem hiding this comment.
Well, rowid virtual column is useful and is used by Modin. I don't want users to 'see' this column when they ask for node size or use an integer index to choose a column. I wrote some examples and it feels really confusing when this virtual column affects size and indexing (e.g. ht[-1] would give you the last column in most cases but for scans, it would be the rowid column). I believe the current Python implementation is the right way to go (we still can reference rowid column by its name). Probably, we should adjust C++ API implementation to have a match.
There was a problem hiding this comment.
Yeah, I think consistent semantics is important. Anyway, it's not related to this PR. LGTM!
No description provided.