Guido van Rossum [Sat, 17 Dec 2022 04:45:55 +0000 (20:45 -0800)]
GH-98831: Add DECREF_INPUTS(), expanding to DECREF() each stack input (#100205)
The presence of this macro indicates that a particular instruction
may be considered for conversion to a register-based format
(see https://github.com/faster-cpython/ideas/issues/485).
An invariant (currently unchecked) is that `DEOPT_IF()` may only
occur *before* `DECREF_INPUTS()`, and `ERROR_IF()` may only occur
*after* it. One reason not to check this is that there are a few
places where we insert *two* `DECREF_INPUTS()` calls, in different
branches of the code. The invariant checking would have to be able
to do some flow control analysis to understand this.
Note that many instructions, especially specialized ones,
can't be converted to use this macro straightforwardly.
This is because the generator currently only generates plain
`Py_DECREF(variable)` statements, and cannot generate
things like `_Py_DECREF_SPECIALIZED()` let alone deal with
`_PyList_AppendTakeRef()`.
Back in commit 226e6e7d4326cf91ef37e13528eb1f62de1bb832
an item was added to the list, renumbering all the rest of the
items, but the forward-reference wasn't updated to match.
Ateeq Sharfuddin [Thu, 15 Dec 2022 22:59:37 +0000 (17:59 -0500)]
gh-78997: AttributeError if loading fails in LibraryLoader.__getattr__
Co-authored-by: Zachary Ware <zachary.ware@gmail.com> Co-authored-by: Filipe Laíns <filipe.lains@gmail.com> Co-authored-by: Eric Wieser <wieser.eric@gmail.com>
Eric Snow [Wed, 14 Dec 2022 18:53:57 +0000 (11:53 -0700)]
gh-90111: Minor Cleanup for Runtime-Global Objects (gh-100254)
* move _PyRuntime.global_objects.interned to _PyRuntime.cached_objects.interned_strings (and use _Py_CACHED_OBJECT())
* rename _PyRuntime.global_objects to _PyRuntime.static_objects
Eric Snow [Mon, 12 Dec 2022 23:51:27 +0000 (16:51 -0700)]
gh-81057: Move _Py_RefTotal to the "Ignored Globals" List (gh-100203)
We can't move it to _PyRuntimeState because the symbol is exposed in the stable ABI. We'll have to sort that out before a per-interpreter GIL, but it shouldn't be too hard.
jarrodcolburn [Mon, 12 Dec 2022 12:25:22 +0000 (06:25 -0600)]
Fix: typo (Indention) (GH-99904)
Example needed to be indented. Was trying to call a context manger `pr` (from ` with cProfile.Profile() as pr:`) wot perform ` pr.print_stats()` once it had already exited.
Gregory P. Smith [Sun, 11 Dec 2022 00:17:39 +0000 (16:17 -0800)]
gh-88500: Reduce memory use of `urllib.unquote` (#96763)
`urllib.unquote_to_bytes` and `urllib.unquote` could both potentially generate `O(len(string))` intermediate `bytes` or `str` objects while computing the unquoted final result depending on the input provided. As Python objects are relatively large, this could consume a lot of ram.
This switches the implementation to using an expanding `bytearray` and a generator internally instead of precomputed `split()` style operations.
Microbenchmarks with some antagonistic inputs like `mess = "\u0141%%%20a%fe"*1000` show this is 10-20% slower for unquote and unquote_to_bytes and no different for typical inputs that are short or lack much unicode or % escaping. But the functions are already quite fast anyways so not a big deal. The slowdown scales consistently linear with input size as expected.
Memory usage observed manually using `/usr/bin/time -v` on `python -m timeit` runs of larger inputs. Unittesting memory consumption is difficult and does not seem worthwhile.
Observed memory usage is ~1/2 for `unquote()` and <1/3 for `unquote_to_bytes()` using `python -m timeit -s 'from urllib.parse import unquote, unquote_to_bytes; v="\u0141%01\u0161%20"*500_000' 'unquote_to_bytes(v)'` as a test.
Eric Snow [Fri, 9 Dec 2022 01:16:37 +0000 (18:16 -0700)]
gh-81057: Fix an ifdef in the time module (#100125)
An earlier commit only defined check_ticks_per_second() when HAVE_TIMES is defined. However, we also need it when HAVE_CLOCK is defined. This primarily affects Windows.
GH-98831: Typed stack effects, and more instructions converted (#99764)
Stack effects can now have a type, e.g. `inst(X, (left, right -- jump/uint64_t)) { ... }`.
Instructions converted to the non-legacy format:
* COMPARE_OP
* COMPARE_OP_FLOAT_JUMP
* COMPARE_OP_INT_JUMP
* COMPARE_OP_STR_JUMP
* STORE_ATTR
* DELETE_ATTR
* STORE_GLOBAL
* STORE_ATTR_INSTANCE_VALUE
* STORE_ATTR_WITH_HINT
* STORE_ATTR_SLOT, and complete the store_attr family
* Complete the store_subscr family: STORE_SUBSCR{,DICT,LIST_INT}
(STORE_SUBSCR was alread half converted,
but wasn't using cache effects yet.)
* DELETE_SUBSCR
* PRINT_EXPR
* INTERPRETER_EXIT (a bit weird, ends in return)
* RETURN_VALUE
* GET_AITER (had to restructure it some)
The original had mysterious `SET_TOP(NULL)` before `goto error`.
I assume those just account for `obj` having been decref'ed,
so I got rid of them in favor of the cleanup implied by `ERROR_IF()`.
* LIST_APPEND (a bit unhappy with it)
* SET_ADD (also a bit unhappy with it)
Ethan Furman [Thu, 8 Dec 2022 06:58:08 +0000 (22:58 -0800)]
gh-100098: [Enum] insist on actual tuples, no subclasses, for auto (GH-100099)
When checking for auto() instances, only top-level usage is supported,
which means either alone or as part of a regular tuple. Other
containers, such as lists, dicts, or namedtuples, will not have auto()
transformed into a value.
Victor Stinner [Wed, 7 Dec 2022 14:22:38 +0000 (15:22 +0100)]
gh-98724: Fix Py_CLEAR() macro side effects (#99100) (#100070)
The Py_CLEAR(), Py_SETREF() and Py_XSETREF() macros now only evaluate
their arguments once. If an argument has side effects, these side
effects are no longer duplicated.
Use temporary variables to avoid duplicating side effects of macro
arguments. If available, use _Py_TYPEOF() to avoid type punning.
Otherwise, use memcpy() for the assignment to prevent a
miscompilation with strict aliasing caused by type punning.
Add _Py_TYPEOF() macro: __typeof__() on GCC and clang.
Add test_py_clear() and test_py_setref() unit tests to _testcapi.
gh-93453: No longer create an event loop in get_event_loop() (#98440)
asyncio.get_event_loop() now always return either running event loop or
the result of get_event_loop_policy().get_event_loop() call. The latter
should now raise an RuntimeError if no current event loop was set
instead of creating and setting a new event loop.
It affects also a number of asyncio functions and constructors which
call get_event_loop() implicitly: ensure_future(), shield(), gather(),
etc.
DeprecationWarning is no longer emitted if there is no running event loop but
the current event loop was set.