Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

gh-102471: convert decimal module to use PyLong_Export API (PEP 757) #128267

Open
wants to merge 4 commits into
base: main
Choose a base branch
from
+35 −23
Open

gh-102471: convert decimal module to use PyLong_Export API (PEP 757) #128267

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
f7f031f
gh-102471: convert decimal module to use PyLong_Export API (PEP 757)
skirpichev Jul 6, 2024
bf687e5
XXX try PyLong_Export without value
skirpichev Dec 26, 2024
021ab8e
Revert "XXX try PyLong_Export without value"
skirpichev Dec 26, 2024
5764614
address review: add asserts
skirpichev Dec 26, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
58 changes: 35 additions & 23 deletions Modules/_decimal/_decimal.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,6 @@
#endif

#include <Python.h>
#include "pycore_long.h" // _PyLong_IsZero()
#include "pycore_pystate.h" // _PyThreadState_GET()
#include "pycore_typeobject.h"
#include "complexobject.h"
Expand Down Expand Up @@ -2323,38 +2322,51 @@ static PyObject *
dec_from_long(decimal_state *state, PyTypeObject *type, PyObject *v,
const mpd_context_t *ctx, uint32_t *status)
{
PyObject *dec;
PyLongObject *l = (PyLongObject *)v;
PyObject *dec = PyDecType_New(state, type);

dec = PyDecType_New(state, type);
if (dec == NULL) {
return NULL;
}

if (_PyLong_IsZero(l)) {
_dec_settriple(dec, MPD_POS, 0, 0);
return dec;
PyLongExport export_long;

if (PyLong_Export(v, &export_long) == -1) {
Py_DECREF(dec);
return NULL;
}
if (export_long.digits) {
const PyLongLayout *layout = PyLong_GetNativeLayout();
const uint32_t base = (uint32_t)1 << layout->bits_per_digit;
const uint8_t sign = export_long.negative ? MPD_NEG : MPD_POS;
const Py_ssize_t len = export_long.ndigits;
Comment on lines +2339 to +2341
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please remove redundant consts.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should not we check if layout->bits_per_digit >= 32?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Please remove redundant consts.

Which one looks redundant for you?

Should not we check if layout->bits_per_digit >= 32?

I was thinking about an assert. Do you think it's ok?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Which one looks redundant for you?

All consts applied to POD types are redundant. 3 lines here and 1 line for export_long.value.

Some style guides require to add const for all variables and function parameters, but this in not in CPython style guide. It makes the code too verbose and more difficult to read.

I was thinking about an assert. Do you think it's ok?

It is okay for now. But if we want to make it an exemplary code, we should remove dependence on CPython implementation details.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

All consts applied to POD types are redundant.

Why? Doing something like value += 1 currently will produce an error at compilation time, while without const - not.

It makes the code too verbose and more difficult to read.

It emphasize programmer intentions not just for compiler, but for human being as well.

It is okay for now.

Ok, then I'll add assertions for all assumptions, that are silent not (endianness, digit order, etc). I think it's fine for a stdlib module. Caching in the module state seems overkill for this.


uint8_t sign = _PyLong_IsNegative(l) ? MPD_NEG : MPD_POS;
assert(layout->bits_per_digit <= 32);
assert(layout->digits_order == -1);
assert(layout->digit_endianness == (PY_LITTLE_ENDIAN ? -1 : 1));
assert(layout->digit_size == 2 || layout->digit_size == 4);

if (_PyLong_IsCompact(l)) {
_dec_settriple(dec, sign, l->long_value.ob_digit[0], 0);
mpd_qfinalize(MPD(dec), ctx, status);
return dec;
if (layout->digit_size == 4) {
mpd_qimport_u32(MPD(dec), export_long.digits, len, sign,
base, ctx, status);
}
else {
mpd_qimport_u16(MPD(dec), export_long.digits, len, sign,
base, ctx, status);
}
PyLong_FreeExport(&export_long);
}
size_t len = _PyLong_DigitCount(l);

#if PYLONG_BITS_IN_DIGIT == 30
mpd_qimport_u32(MPD(dec), l->long_value.ob_digit, len, sign, PyLong_BASE,
ctx, status);
#elif PYLONG_BITS_IN_DIGIT == 15
mpd_qimport_u16(MPD(dec), l->long_value.ob_digit, len, sign, PyLong_BASE,
ctx, status);
#else
#error "PYLONG_BITS_IN_DIGIT should be 15 or 30"
#endif
else {
const int64_t value = export_long.value;

if (-(int64_t)UINT32_MAX <= value && value <= (int64_t)UINT32_MAX) {
_dec_settriple(dec, value < 0 ? MPD_NEG : MPD_POS,
(uint32_t)Py_ABS(value), 0);
mpd_qfinalize(MPD(dec), ctx, status);
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is it worth it to have this fast-path? (Why not always calling mpd_qset_i64()?)

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's up to reviewers. In old pr I did benchmarks:
#127925 (comment)

}
else {
mpd_qset_i64(MPD(dec), value, ctx, status);
}
}
return dec;
}

Expand Down
Loading