ENH: allow clients to send extra compute and odo kwargs #1342
Conversation
llllllllll
force-pushed
the
server-extra-args
branch
2 times, most recently
from
19a49db
to
5161ad9
Compare
blaze.compute accepts keyword args that can change how computations are performed. This change forwards two extra dictionaries to the server for use in computation and odo conversion. These are: compute_kwargs - **unpacked into the compute call on the server odo_kwargs - **unpacked into the odo call on the server
llllllllll
force-pushed
the
server-extra-args
branch
from
5161ad9
to
ea2f758
Compare
| try: | ||
| return typ(x) | ||
| except TypeError: | ||
| return odo(x, typ) | ||
| return odo(x, typ, **odo_kwargs or {}) |
There was a problem hiding this comment.
I presume the precedence here is equivalent to **(odo_kwargs or {}), correct? What about putting explicit parens around the disjunction to make it more clear -- the spacing makes it look strange at first read.
|
Updated to address the comments, anything else before merging? |
| good_query = { | ||
| 'expr': to_tree(expr), | ||
| 'compute_kwargs': { | ||
| 'return_df': odo(DumbResource.df, list), |
There was a problem hiding this comment.
There's a subtlety here -- this is the same kv pair as for the odo_kwargs above which I didn't see at first; since it ends up hitting the compute_down() dispatch on line 62, it ultimately calls dumb_to_df and everything works out.
There was a problem hiding this comment.
Correct. In the odo_kwargs test, expr is just a field which does not call compute_down. This changes the expr to have a sort on it causing it hit the compute_down and that is where the resource is converted into a dataframe.
|
@llllllllll LGTM, thanks. |
blaze.computeaccepts keyword args that can change how computations areperformed. This change forwards two extra dictionaries to the server for
use in computation and odo conversion. These are:
compute_kwargs-**unpacked into the compute call on the serverodo_kwargs-**unpacked into the odo call on the server