URL |
Status |
Linked |
Edit |
PR 29404 |
merged |
iritkatriel,
2021-11-04 13:24
|
|
PR 29495 |
merged |
iritkatriel,
2021-11-09 17:32
|
|
PR 29518 |
merged |
iritkatriel,
2021-11-10 19:17
|
|
PR 29528 |
merged |
brandtbucher,
2021-11-11 23:41
|
|
PR 29627 |
merged |
iritkatriel,
2021-11-18 19:25
|
|
PR 29780 |
merged |
iritkatriel,
2021-11-25 17:45
|
|
PR 29890 |
merged |
iritkatriel,
2021-12-02 12:37
|
|
PR 29922 |
merged |
iritkatriel,
2021-12-05 17:31
|
|
PR 30122 |
merged |
iritkatriel,
2021-12-15 14:40
|
|
PR 30289 |
open |
iritkatriel,
2022-01-01 14:03
|
|
PR 30849 |
merged |
iritkatriel,
2022-01-24 10:29
|
|
msg405681 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-04 11:29 |
Exceptions are represented in the interpreter as (type, val, tb) triplets which most of the time contain redundant information (the type is the type of val and the tb is also on the exception). This complicates the code and is inefficient as opcodes that manage exceptions push and pop 3 items for each exception.
We will change the internal representation to be (1) just the exception value if it is normalised and (2) a tuple of the 3 values for the uncommon case where they are all needed.
See also https://github.com/faster-cpython/ideas/issues/106.
|
msg405839 - (view) |
Author: Terry J. Reedy (terry.reedy) * |
Date: 2021-11-06 01:34 |
Would there be any change at the Python level?
|
msg405849 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-06 07:53 |
Initially not, neither in python nor in the c api.
It would be nice to replace PyErr_Fetch/Restore by a version that takes just an exception but that’s a long deprecation.
|
msg406114 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-10 16:57 |
New changeset 05fbd60147456d77a7aecf29dddd86c5bde5872f by Irit Katriel in branch 'main':
bpo-45711: Use _PyErr_ClearExcState instead of setting only exc_value to NULL (GH-29404)
https://github.com/python/cpython/commit/05fbd60147456d77a7aecf29dddd86c5bde5872f
|
msg406120 - (view) |
Author: Mark Shannon (Mark.Shannon) * |
Date: 2021-11-10 18:08 |
New changeset 4cdeee5978ee3f8ea7fe95172ae04d866cd88177 by Irit Katriel in branch 'main':
bpo-45711: remove unnecessary DUP_TOP and POP in exception handling (GH-29495)
https://github.com/python/cpython/commit/4cdeee5978ee3f8ea7fe95172ae04d866cd88177
|
msg406209 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-12 13:21 |
New changeset 8f1b71de731dda668aede7c9b34d0ad7afb8f6a8 by Brandt Bucher in branch 'main':
bpo-45711: Re-bump the magic number and update doc (GH-29528)
https://github.com/python/cpython/commit/8f1b71de731dda668aede7c9b34d0ad7afb8f6a8
|
msg406238 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-12 21:37 |
New changeset de3db1448b1b983eeb9f4498d07e3d2f1fb6d29d by Irit Katriel in branch 'main':
bpo-45711: assert that the type of exc_info is redundant (GH-29518)
https://github.com/python/cpython/commit/de3db1448b1b983eeb9f4498d07e3d2f1fb6d29d
|
msg406983 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-25 09:41 |
New changeset c456dfafe9f9f6614fbcf2213a93707f0e101f4e by Irit Katriel in branch 'main':
bpo-45711: use exc_value instead of exc_type to determine if exc_info is valid. Add more assertions. (GH-29627)
https://github.com/python/cpython/commit/c456dfafe9f9f6614fbcf2213a93707f0e101f4e
|
msg406984 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-25 09:53 |
Following the analysis/discussion on https://github.com/faster-cpython/ideas/issues/106:
* exc_info is always normalized, so we actually will never need to create a tuple of these three values (exc_curinfo, the exception raised but not yet caught can be unnormalized, but it is not pushed/popped on the stack).
* We will reduce the interpreter's exc_info representation to just the exception instance.
* There are two APIs that are impacted, both in non-documented edge cases:
1. sys.exc_info()[2] can currently be different from sys.exc_info()[1].__traceback__ because changes to the latter (while an except clause is executing) don't show up in the former. This is arguably a bug, we will change it so that the type and traceback are always consistent with the exception.
2. PyErr_SetExcInfo does no arg checking, and will set exc_info to an inconsistent triplet if you ask it to. However, the exc_value arg must be an exception instance so the only thing you can do is pass in nonsensical args where the type/traceback do not match the exception. This function's purpose is to save/restore exc_info. We will make it ignore the type and traceback and document that change.
|
msg407402 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-11-30 22:37 |
New changeset 8a45ca542a65ea27e7acaa44a4c833a27830e796 by Irit Katriel in branch 'main':
bpo-45711: Change exc_info related APIs to derive type and traceback from the exception instance (GH-29780)
https://github.com/python/cpython/commit/8a45ca542a65ea27e7acaa44a4c833a27830e796
|
msg407597 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-12-03 19:05 |
New changeset 2ff758bd1a144ee712e96ae1db91f476c3b252bb by Irit Katriel in branch 'main':
bpo-45711: [asyncio] Normalize exceptions immediately after Fetch, before they are stored as StackItem, which should be normalized (GH-29890)
https://github.com/python/cpython/commit/2ff758bd1a144ee712e96ae1db91f476c3b252bb
|
msg408043 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2021-12-08 20:08 |
New changeset 2109f7880b65755329a877da3a7f8a362de07350 by Irit Katriel in branch 'main':
bpo-45711: Remove unnecessary normalization of exc_info (GH-29922)
https://github.com/python/cpython/commit/2109f7880b65755329a877da3a7f8a362de07350
|
msg408768 - (view) |
Author: Stefan Behnel (scoder) * |
Date: 2021-12-17 10:45 |
FYI, we track the Cython side of this in
https://github.com/cython/cython/issues/4500
|
msg408786 - (view) |
Author: Mark Shannon (Mark.Shannon) * |
Date: 2021-12-17 14:46 |
New changeset 396b58345f81d4c8c5a52546d2288e666a1b9b8b by Irit Katriel in branch 'main':
bpo-45711: Remove type and traceback from exc_info (GH-30122)
https://github.com/python/cpython/commit/396b58345f81d4c8c5a52546d2288e666a1b9b8b
|
msg409453 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-01-01 10:37 |
This is done, there will be more cleanups that this enables but they can have their own issues.
|
msg411497 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-01-24 17:45 |
New changeset 80e1def9ded2a1d017410394e50c88aa39135029 by Irit Katriel in branch 'main':
bpo-45711: move whatsnew entries which are incorrectly listed under New Features (GH-30849)
https://github.com/python/cpython/commit/80e1def9ded2a1d017410394e50c88aa39135029
|
msg412264 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-02-01 14:19 |
Would it be possible to revert the change until Cython and numpy are ready for it?
> bpo-45711: Remove type and traceback from exc_info (GH-30122)
This change broke numpy which uses C code generated by Cython for coroutines in numpy/random/_mt19937.c (file generated from numpy/random/_mt19937.pyx).
Example of Cython code which no fails with a compiler error since exc_info->exc_type has been removed:
/* GetTopmostException */
#if CYTHON_USE_EXC_INFO_STACK
static _PyErr_StackItem *
__Pyx_PyErr_GetTopmostException(PyThreadState *tstate)
{
_PyErr_StackItem *exc_info = tstate->exc_info;
while ((exc_info->exc_type == NULL || exc_info->exc_type == Py_None) &&
exc_info->previous_item != NULL)
{
exc_info = exc_info->previous_item;
}
return exc_info;
}
#endif
I propose this plan:
* (1) Revert this specific change
* (2) Wait until a change is merged in Cython 0.29.x
* (3) Wait until a new Cython 0.29.x version is released
* (4) Wait until numpy is released with regenerated code compatible with this change
* (5) Apply again the change
It should increase the number of packages when Python 3.11 will be released.
In Cython, the issue is tracked as: https://github.com/cython/cython/issues/4500
|
msg412265 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 14:28 |
I can probably just put back the two fields in _PyErr_StackItem and make sure they get updated when exc_value is set.
Reverting the whole thing would include big changes in compile.c and ceval.c, so I'd rather avoid that if we can.
|
msg412270 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 14:52 |
Does unsetting CYTHON_USE_EXC_INFO_STACK still work?
#if CYTHON_USE_EXC_INFO_STACK
// See https://bugs.python.org/issue25612
#define __Pyx_ExcInfoStruct _PyErr_StackItem
#else
// Minimal replacement struct for Py<3.7, without the Py3.7 exception state stack.
typedef struct {
PyObject *exc_type;
PyObject *exc_value;
PyObject *exc_traceback;
} __Pyx_ExcInfoStruct;
#endif
|
msg412272 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-02-01 15:07 |
> In Cython, the issue is tracked as: https://github.com/cython/cython/issues/4500
I don't understand the Cython status. The code was updated in the master branch:
* https://github.com/cython/cython/pull/4584
* https://github.com/cython/cython/commit/776957022d062ed24edea192b719720118ee3576
https://github.com/cython/cython/issues/4500 was closed even if the change was not backported to 0.29.x.
I understand that there is no released Cython 0.29.x version compatible with this change.
|
msg412273 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-02-01 15:12 |
I commented the Cython issue:
https://github.com/cython/cython/issues/4500#issuecomment-1026949365
I also opened a discussion on python-dev: "Please update Cython before introcuding C API incompatible changes in Python".
https://mail.python.org/archives/list/python-dev@python.org/thread/RS2C53LDZPXHRR2VCY2G2YSPDVA4LNQU/
|
msg412275 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-02-01 15:19 |
Irit Katriel:
> Reverting the whole thing would include big changes in compile.c and ceval.c, so I'd rather avoid that if we can.
It was a good idea to split changes of this issue into multiple commits :-) I'm only proposing to revert the one which introduces the C API incompatible changes: the commit 396b58345f81d4c8c5a52546d2288e666a1b9b8b "bpo-45711: Remove type and traceback from exc_info (GH-30122)".
Maybe Cython can be quickly updated, but the situation seems to be complicated :-( https://github.com/cython/cython/issues/4500
|
msg412281 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 15:47 |
That commit has significant changes in ceval.c and compile.c. They don't need to be reverted to unbreak cython. I'm working on a PR for a simpler change.
Have you tried with CYTHON_USE_EXC_INFO_STACK undefined?
|
msg412289 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 16:46 |
If this is still the position of cython maintainers:
https://github.com/cython/cython/issues/4581#issuecomment-1016503683
then I will need to revert the change until 3.12.
|
msg412290 - (view) |
Author: Guido van Rossum (gvanrossum) * |
Date: 2022-02-01 16:50 |
Time to insist on directly communicating with the Cython team (esp. @scoder) and broker some kind of compromise.
|
msg412292 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-02-01 17:08 |
> __Pyx_PyErr_GetTopmostException(PyThreadState *tstate)
Python provides a *private* _PyErr_GetTopmostException(tstate) function, but Cython reimplements its own function. I'm not sure why.
GH-30531 proposes adding PyErr_GetActiveException() function which has no parameter, but Cython __Pyx_PyErr_GetTopmostException() has a tstate parameter.
Simplified example numpy/random/_mt19937.c code:
static CYTHON_INLINE void
__Pyx__ExceptionSave(PyThreadState *tstate, PyObject **type,
PyObject **value, PyObject **tb)
{
_PyErr_StackItem *exc_info = __Pyx_PyErr_GetTopmostException(tstate);
*type = exc_info->exc_type;
*value = exc_info->exc_value;
*tb = exc_info->exc_traceback;
Py_XINCREF(*type);
Py_XINCREF(*value);
Py_XINCREF(*tb);
}
static CYTHON_INLINE void
__Pyx__ExceptionReset(PyThreadState *tstate, PyObject *type,
PyObject *value, PyObject *tb)
{
PyObject *tmp_type, *tmp_value, *tmp_tb;
_PyErr_StackItem *exc_info = tstate->exc_info;
tmp_type = exc_info->exc_type;
tmp_value = exc_info->exc_value;
tmp_tb = exc_info->exc_traceback;
exc_info->exc_type = type;
exc_info->exc_value = value;
exc_info->exc_traceback = tb;
Py_XDECREF(tmp_type);
Py_XDECREF(tmp_value);
Py_XDECREF(tmp_tb);
}
Cython saves/restores the current exception of tstate. Maybe we need to provide a high-level API for that as well?
|
msg412303 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 19:32 |
This is a backport of @scoder's patch to 0.29.x. (I don't know if this is helpful).
https://github.com/cython/cython/compare/master...iritkatriel:exc_info?expand=1
|
msg412304 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-02-01 19:52 |
> GH-30531 proposes adding PyErr_GetActiveException() function which has no parameter, but Cython __Pyx_PyErr_GetTopmostException() has a tstate parameter.
I've now updated it to follow the pattern of other functions, where the is a private function that takes tstate and the public function calls it.
So it adds in Include/pyerrors.h
PyAPI_FUNC(PyObject*) PyErr_GetActiveException(void);
PyAPI_FUNC(void) PyErr_SetActiveException(PyObject *);
and in Include/cpython/pyerrors.h
PyAPI_FUNC(PyObject*) _PyErr_GetActiveException(PyThreadState *);
PyAPI_FUNC(void) _PyErr_SetActiveException(PyThreadState *, PyObject *);
|
msg412316 - (view) |
Author: (da-woods) * |
Date: 2022-02-01 22:02 |
Just a quick comment on Cython and these changes:
Cython 0.29 can build itself on Python 3.11a4 with `CFLAGS="-DCYTHON_FAST_THREAD_STATE=0 -DCYTHON_USE_EXC_INFO_STACK=0" python3.11 setup.py build_ext`.
I think there's some coroutine code left that isn't fixed by those flags, but a large chunk of Cython libraries won't use coroutines and so will work fine with these flags. Hopefully that unblocks some stuff for you.
|
msg414482 - (view) |
Author: STINNER Victor (vstinner) * |
Date: 2022-03-03 22:14 |
lxml does crash on the current Cython 0.29.x development branch.
|
msg415086 - (view) |
Author: Irit Katriel (iritkatriel) * |
Date: 2022-03-13 19:33 |
We have agreed on python-dev [1] that the cpython changes will not be reverted, and the issue will be fixed in cython. So I am closing this again.
[1] https://mail.python.org/archives/list/python-dev@python.org/message/BHIQL4P6F7OPMCAP6U24XEZUPQKI62UT/
|
|
Date |
User |
Action |
Args |
2022-04-11 14:59:52 | admin | set | github: 89874 |
2022-03-13 19:33:16 | iritkatriel | set | status: open -> closed resolution: fixed messages:
+ msg415086
|
2022-03-03 22:14:23 | vstinner | set | messages:
+ msg414482 |
2022-02-01 22:02:02 | da-woods | set | nosy:
+ da-woods messages:
+ msg412316
|
2022-02-01 19:52:16 | iritkatriel | set | messages:
+ msg412304 |
2022-02-01 19:32:23 | iritkatriel | set | messages:
+ msg412303 |
2022-02-01 17:08:38 | vstinner | set | messages:
+ msg412292 |
2022-02-01 16:50:45 | gvanrossum | set | messages:
+ msg412290 |
2022-02-01 16:46:59 | iritkatriel | set | messages:
+ msg412289 |
2022-02-01 15:47:03 | iritkatriel | set | messages:
+ msg412281 |
2022-02-01 15:19:00 | vstinner | set | messages:
+ msg412275 |
2022-02-01 15:12:45 | vstinner | set | messages:
+ msg412273 |
2022-02-01 15:07:35 | vstinner | set | messages:
+ msg412272 |
2022-02-01 14:52:41 | iritkatriel | set | messages:
+ msg412270 |
2022-02-01 14:28:12 | iritkatriel | set | messages:
+ msg412265 |
2022-02-01 14:19:59 | vstinner | set | status: closed -> open
nosy:
+ vstinner messages:
+ msg412264
resolution: fixed -> (no value) |
2022-01-24 17:45:15 | iritkatriel | set | messages:
+ msg411497 |
2022-01-24 10:29:59 | iritkatriel | set | pull_requests:
+ pull_request29031 |
2022-01-01 14:03:04 | iritkatriel | set | pull_requests:
+ pull_request28539 |
2022-01-01 10:37:51 | iritkatriel | set | pull_requests:
- pull_request28217 |
2022-01-01 10:37:40 | iritkatriel | set | pull_requests:
- pull_request28193 |
2022-01-01 10:37:29 | iritkatriel | set | status: open -> closed resolution: fixed messages:
+ msg409453
stage: patch review -> resolved |
2022-01-01 10:36:19 | iritkatriel | set | pull_requests:
- pull_request28503 |
2021-12-29 14:37:17 | iritkatriel | set | pull_requests:
+ pull_request28503 |
2021-12-17 14:46:33 | Mark.Shannon | set | messages:
+ msg408786 |
2021-12-17 10:45:39 | scoder | set | messages:
+ msg408768 |
2021-12-17 09:49:45 | scoder | set | nosy:
+ scoder
|
2021-12-15 14:40:12 | iritkatriel | set | pull_requests:
+ pull_request28341 |
2021-12-08 22:31:08 | iritkatriel | set | pull_requests:
+ pull_request28217 |
2021-12-08 20:08:11 | iritkatriel | set | messages:
+ msg408043 |
2021-12-07 17:47:35 | iritkatriel | set | pull_requests:
+ pull_request28193 |
2021-12-05 17:31:12 | iritkatriel | set | pull_requests:
+ pull_request28146 |
2021-12-03 19:05:23 | iritkatriel | set | messages:
+ msg407597 |
2021-12-02 12:37:21 | iritkatriel | set | pull_requests:
+ pull_request28114 |
2021-11-30 22:37:29 | iritkatriel | set | messages:
+ msg407402 |
2021-11-25 17:45:45 | iritkatriel | set | pull_requests:
+ pull_request28017 |
2021-11-25 09:53:52 | iritkatriel | set | messages:
+ msg406984 |
2021-11-25 09:41:41 | iritkatriel | set | messages:
+ msg406983 |
2021-11-18 19:25:08 | iritkatriel | set | pull_requests:
+ pull_request27859 |
2021-11-12 21:37:36 | iritkatriel | set | messages:
+ msg406238 |
2021-11-12 13:21:53 | iritkatriel | set | messages:
+ msg406209 |
2021-11-11 23:41:36 | brandtbucher | set | nosy:
+ brandtbucher pull_requests:
+ pull_request27778
|
2021-11-10 19:17:44 | iritkatriel | set | pull_requests:
+ pull_request27770 |
2021-11-10 18:08:41 | Mark.Shannon | set | messages:
+ msg406120 |
2021-11-10 16:57:23 | iritkatriel | set | messages:
+ msg406114 |
2021-11-09 17:32:00 | iritkatriel | set | pull_requests:
+ pull_request27745 |
2021-11-06 07:53:27 | iritkatriel | set | messages:
+ msg405849 |
2021-11-06 01:34:06 | terry.reedy | set | type: performance
messages:
+ msg405839 nosy:
+ terry.reedy |
2021-11-04 13:24:54 | iritkatriel | set | keywords:
+ patch stage: patch review pull_requests:
+ pull_request27660 |
2021-11-04 11:29:47 | iritkatriel | create | |