30519033396

Committed 30 Jul 2026 06:15AM UTC coverage: 90.546% (-3.2%) from 93.725%

Build # 30519033396

Build Type

push

github

Committed by

web-flow

Commit Message

CI: Measure Numba JIT coverage in a single dedicated job (#913)

Remove the five Numba decorator patterns (@jit, @njit, @overload,
@guvectorize) from .coveragerc exclude_lines. coverage.py excludes the
entire decorated function when a decorator line matches, so ~1,700
statements of jitted code were invisible to coverage. Also omit the
test tree (*/tests/*) so the headline figure measures library code
only.

Since numba 0.61 (fixed to respect coverage configuration in 0.62), the
compiler emits coverage for JIT functions at compile time when
NUMBA_JIT_COVERAGE=1. Because emission happens during lowering, functions
with cache=True loaded from a warm on-disk cache report nothing, so the
coverage job points NUMBA_CACHE_DIR at a throwaway directory to force
compilation.

Restructure ci.yml accordingly:

- The OS x Python matrix now runs plain pytest (the production
  configuration, no coverage overhead).
- A new single `coverage` job (ubuntu, Python 3.13) runs the suite once
  with NUMBA_JIT_COVERAGE=1 and uploads to Coveralls. With one producer,
  the old per-matrix Coveralls gate is obsolete.
- Pin numba>=0.62 in environment.yml; without the env var taking effect
  the reported total would silently drop by ~5 pp.

New baseline: 90.55% (5,638 source statements, 533 missed), 600 tests
passing.

Closes #871

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

Coverage Stats

5105 of 5638 relevant lines covered (90.55%)

0.91 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

92.86

/quantecon/util/array.py

"""
Array Utilities
===============

Array
-----
searchsorted

"""

import warnings
from numba import jit, objmode

# ----------------- #
# -ARRAY UTILITIES- #
# ----------------- #

@jit(nopython=True)
def searchsorted(a, v):
    """
    Custom version of np.searchsorted. Return the largest index `i` such
    that `a[i-1] <= v < a[i]` (for `i = 0`, `v < a[0]`); if `v[n-1] <=
    v`, return `n`, where `n = len(a)`.

    .. deprecated::

        Deprecated, use `np.searchsorted(a, v, side='right')` instead.

    Parameters
    ----------
    a : ndarray(float, ndim=1)
        Input array. Must be sorted in ascending order.

    v : scalar(float)
        Value to be compared with elements of `a`.

    Returns
    -------
    scalar(int)
        Largest index `i` such that `a[i-1] <= v < a[i]`, or len(a) if
        no such index exists.

    Notes
    -----
    This routine is jit-complied if the module Numba is vailable; if
    not, it is an alias of np.searchsorted(a, v, side='right').

    Examples
    --------
    >>> a = np.array([0.2, 0.4, 1.0])
    >>> searchsorted(a, 0.1)
    0
    >>> searchsorted(a, 0.4)
    2
    >>> searchsorted(a, 2)
    3

    """
    with objmode():
        warnings.warn(
            "`searchsorted(a, v)` is deprecated. "
            "Use `np.searchsorted(a, v, side='right')` instead.",
            DeprecationWarning,
            stacklevel=2,
        )

    lo = -1
    hi = len(a)
    while(lo < hi-1):
        m = (lo + hi) // 2
        if v < a[m]:
            hi = m
        else:
            lo = m
    return hi

1	"""
2	Array Utilities
3	===============
4
5	Array
6	-----
7	searchsorted
8
9	"""
10
11	import warnings	1✔
12	from numba import jit, objmode	1✔
13
14	# ----------------- #
15	# -ARRAY UTILITIES- #
16	# ----------------- #
17
18	@jit(nopython=True)	1✔
19	def searchsorted(a, v):	1✔
20	"""
21	Custom version of np.searchsorted. Return the largest index `i` such
22	that `a[i-1] <= v < a[i]` (for `i = 0`, `v < a[0]`); if `v[n-1] <=
23	v`, return `n`, where `n = len(a)`.
24
25	.. deprecated::
26
27	Deprecated, use `np.searchsorted(a, v, side='right')` instead.
28
29	Parameters
30	----------
31	a : ndarray(float, ndim=1)
32	Input array. Must be sorted in ascending order.
33
34	v : scalar(float)
35	Value to be compared with elements of `a`.
36
37	Returns
38	-------
39	scalar(int)
40	Largest index `i` such that `a[i-1] <= v < a[i]`, or len(a) if
41	no such index exists.
42
43	Notes
44	-----
45	This routine is jit-complied if the module Numba is vailable; if
46	not, it is an alias of np.searchsorted(a, v, side='right').
47
48	Examples
49	--------
50	>>> a = np.array([0.2, 0.4, 1.0])
51	>>> searchsorted(a, 0.1)
52	0
53	>>> searchsorted(a, 0.4)
54	2
55	>>> searchsorted(a, 2)
56	3
57
58	"""
59	with objmode():	1✔
60	warnings.warn(	×
61	"`searchsorted(a, v)` is deprecated. "
62	"Use `np.searchsorted(a, v, side='right')` instead.",
63	DeprecationWarning,
64	stacklevel=2,
65	)
66
67	lo = -1	1✔
68	hi = len(a)	1✔
69	while(lo < hi-1):	1✔
70	m = (lo + hi) // 2	1✔
71	if v < a[m]:	1✔
72	hi = m	1✔
73	else:
74	lo = m	1✔
75	return hi	1✔

QuantEcon / QuantEcon.py / 30519033396

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous