doc: some performance notes for query_range
This commit is contained in:
parent
ad55c5c345
commit
0f9eca5722
2 changed files with 6 additions and 1 deletions
|
@ -538,6 +538,9 @@ def query_hpi_functions(
|
|||
# chain list of functions from user, in the order they wrote them on the CLI
|
||||
input_src = chain(*(f() for f in _locate_functions_or_prompt(qualified_names)))
|
||||
|
||||
# NOTE: if passing just one function to this which returns a single namedtuple/dataclass,
|
||||
# using --order-key will often be faster as it does not need to duplicate the iterator in memory,
|
||||
# or try to find the --order-value type on each object before sorting
|
||||
res = select_range(
|
||||
input_src,
|
||||
order_key=order_key,
|
||||
|
|
|
@ -337,6 +337,8 @@ def select_range(
|
|||
# if the user supplied a order_key, and/or we've generated an order_value, create
|
||||
# the function that accesses that type on each value in the iterator
|
||||
if order_key is not None or order_value is not None:
|
||||
# _generate_order_valeu_func internally here creates a copy of the iterator, which has to
|
||||
# be consumed in-case we're sorting by mixed types
|
||||
order_by_chosen, itr = _handle_generate_order_by(itr, order_key=order_key, order_value=order_value)
|
||||
# signifies that itr is empty -- can early return here
|
||||
if order_by_chosen is None:
|
||||
|
@ -398,7 +400,7 @@ Specify a type or a key to order the value by""")
|
|||
return itr
|
||||
|
||||
|
||||
# re-use items from query for testing
|
||||
# reuse items from query for testing
|
||||
from .query import _A, _B, _Float, _mixed_iter_errors
|
||||
|
||||
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue