[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Change by Erlend Egeberg Aasland : -- pull_requests: +22956 pull_request: https://github.com/python/cpython/pull/24126 ___ Python tracker <https://bugs.python.org/issue41

[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Erlend Egeberg Aasland added the comment: > I'm also in the process of removing PySocketModuleAPI from _ssl.c. BTW, do we still want to keep the socket CAPI for external users? -- ___ Python tracker <https://bugs.python.org/issu

[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Change by Erlend Egeberg Aasland : -- pull_requests: +22947 pull_request: https://github.com/python/cpython/pull/24117 ___ Python tracker <https://bugs.python.org/issue41

[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Erlend Egeberg Aasland added the comment: > I'm also in the process of removing PySocketModuleAPI from _ssl.c. That's the only user of the API, right? In that case, I'll target _decimal instead. Hai Shi is working on _curses and _datetime, AF

[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Erlend Egeberg Aasland added the comment: > Welcome to join this bpo, you can enhance those extension modules if you like > :) Thanks :) I'll take a look at socket first. -- ___ Python tracker <https://bugs.python.org/i

[issue41798] [C API] Revisit usage of the PyCapsule C API with multi-phase initialization API

Erlend Egeberg Aasland added the comment: Hai Shi, do you mind if I have a go at the socket module CAPI capsule? -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue41

[issue40810] sqlite3 test CheckTraceCallbackContent fails for sqlite v3.7.3 through 3.7.14.1

Erlend Egeberg Aasland added the comment: > +1 for dropping support for < 3.7.15 in master. We should fix or skip tests > in maintenance branches. PR's for fixing the tests in 3.9 and 3.8 are now awaiting merge. PR for dropping support for SQLite pre 3.7.15 in master

[issue40810] sqlite3 test CheckTraceCallbackContent fails for sqlite v3.7.3 through 3.7.14.1

Change by Erlend Egeberg Aasland : -- pull_requests: +22935 pull_request: https://github.com/python/cpython/pull/24106 ___ Python tracker <https://bugs.python.org/issue40

[issue40822] Drop support for SQLite pre 3.7.15

Change by Erlend Egeberg Aasland : -- stage: -> resolved status: open -> closed ___ Python tracker <https://bugs.python.org/issue40822> ___ ___ Pyth

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Erlend Egeberg Aasland added the comment: > FWIW, any core dev can merge SQLite updates for Windows (if they're satisfied > there are no issues). A CI check would indicate that there are no issues. The PoC CI is just a short bash script wrapped in GitHub's workflow syntax. I'd be

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Change by Erlend Egeberg Aasland : -- pull_requests: +22926 pull_request: https://github.com/python/cpython/pull/24095 ___ Python tracker <https://bugs.python.org/issue1635

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Change by Erlend Egeberg Aasland : Added file: https://bugs.python.org/file49718/0001-Add-simple-validation-CI.patch ___ Python tracker <https://bugs.python.org/issue42

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Change by Erlend Egeberg Aasland : Removed file: https://bugs.python.org/file49694/0001-Add-simple-validation-CI.patch ___ Python tracker <https://bugs.python.org/issue42

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Erlend Egeberg Aasland added the comment: Ok, perhaps we should leave a comment about that? It already has a comment about multi-phase init. What about Modules/_testbuffer.c and the xx-modules? $ grep -E 'static PyTypeObject .* =' $(find . -name "*.c"|grep -vE '(Do

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22899 pull_request: https://github.com/python/cpython/pull/24066 ___ Python tracker <https://bugs.python.org/issue40

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22898 pull_request: https://github.com/python/cpython/pull/24065 ___ Python tracker <https://bugs.python.org/issue40

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Erlend Egeberg Aasland added the comment: Should Modules/_testcapimodule.c stay untouched? -- ___ Python tracker <https://bugs.python.org/issue40077> ___ ___

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22888 stage: -> patch review pull_request: https://github.com/python/cpython/pull/24053 ___ Python tracker <https://bugs.python.org/issu

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: Yes, R. Hipp has fixed this here: https://www2.sqlite.org/cgi/src/info/e5d7209e118a8453 We still need to modify PCbuild in order to build the SQLite library with math functions from 3.35.0 and onwards

[issue38780] SysLogHandler crash atexit

Erlend Egeberg Aasland added the comment: Using a NullSocket fixes the test in commit f0f8ff8f06afd43947e268824aa4381add05ce8f -- ___ Python tracker <https://bugs.python.org/issue38

[issue38780] SysLogHandler crash atexit

Erlend Egeberg Aasland added the comment: I've forked your work, Jason: https://github.com/erlend-aasland/cpython/tree/bpo-38780-test I added the test from msg356475, and I'm able to reproduce it by asserting that the `socket` attribute is present (commit

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: FYI, I've opened a thread on the SQLite Forum: https://sqlite.org/forum/forumpost/721645dc36 -- ___ Python tracker <https://bugs.python.org/issue42

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: Using SQLite 3.35.0 nightly build, building Python using PCbuild/build.bat and PCBuild\pcbuild.sln with VS 2017 (does this reflect how the Windows team build the installers?), I can confirm that math functions are _not_ included by default. I added

[issue24464] "sqlite3_enable_shared_cache" deprecation warning when compiling with macOS system SQLite3

Change by Erlend Egeberg Aasland : -- pull_requests: +22850 pull_request: https://github.com/python/cpython/pull/24008 ___ Python tracker <https://bugs.python.org/issue24

[issue40956] Use Argument Clinic in sqlite3

Change by Erlend Egeberg Aasland : -- pull_requests: +22849 pull_request: https://github.com/python/cpython/pull/24007 ___ Python tracker <https://bugs.python.org/issue40

[issue42393] Raise overflow errors iso. deprecation warnings in socket.htons and socket.ntohs (was deprecated in 3.7)

Change by Erlend Egeberg Aasland : -- pull_requests: +22825 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23980 ___ Python tracker <https://bugs.python.org/issu

[issue42243] Don't access the module dictionary directly

Change by Erlend Egeberg Aasland : -- stage: patch review -> resolved status: open -> closed ___ Python tracker <https://bugs.python.org/issue42243> ___ ___

[issue42243] Don't access the module dictionary directly

Change by Erlend Egeberg Aasland : -- resolution: -> wont fix ___ Python tracker <https://bugs.python.org/issue42243> ___ ___ Python-bugs-list mailing list Un

[issue42048] Document Argument Clinic's defining_class converter

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22823 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23978 ___ Python tracker <https://bugs.python.org/issu

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22819 pull_request: https://github.com/python/cpython/pull/23975 ___ Python tracker <https://bugs.python.org/issue40

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Change by Erlend Egeberg Aasland : -- pull_requests: +22817 pull_request: https://github.com/python/cpython/pull/23972 ___ Python tracker <https://bugs.python.org/issue1635

[issue42755] sqlite3.Connection.backup default value is diffrent between implmentation and docs

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22810 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23965 ___ Python tracker <https://bugs.python.org/issu

[issue40956] Use Argument Clinic in sqlite3

Change by Erlend Egeberg Aasland : -- pull_requests: +22809 pull_request: https://github.com/python/cpython/pull/23964 ___ Python tracker <https://bugs.python.org/issue40

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: > So, can this be closed? For macOS, the SQLITE_ENABLE_MATH_FUNCTIONS option is enabled by default, but I'm not sure about the Windows build. The SQLite docs says: "This option is also included on Windows builds using the "Makefile.m

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: Seems like the math functions will be included by default anyways, so we'll have to explicitly exclude them if we don't want them. https://www.sqlite.org/draft/compile.html#enable_math_functions: "This option [SQLITE_ENABLE_MATH_FUNC

[issue42686] include built-in Math functions in SQLite to 3.35.0 of march 2021

Erlend Egeberg Aasland added the comment: I see no reason not to build SQLite with -DSQLITE_ENABLE_MATH_FUNCTIONS for the Windows and macOS builds. Why take the detour through the Python library when you can use the built-in SQLite functions? The change is trivial: Add

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Erlend Egeberg Aasland added the comment: See attached proof-of-concept CI check. This GitHub Workflow extracts the SQLite version information, downloads the amalgamation source code, and does a simple `cmp` on each file. Example run: https://github.com/erlend-aasland/cpython-source-deps

[issue40866] Use PyModule_AddType() in posix module initialisation

Change by Erlend Egeberg Aasland : -- stage: patch review -> resolved status: open -> closed ___ Python tracker <https://bugs.python.org/issue40866> ___ ___

[issue42624] sqlite3 package document mistake

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue42624> ___ ___ Python-bugs-list mailing list Unsub

[issue24139] Use sqlite3 extended error codes

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue24139> ___ ___ Python-bugs-list mailing list Unsub

[issue40956] Use Argument Clinic in sqlite3

Change by Erlend Egeberg Aasland : -- pull_requests: +22697 pull_request: https://github.com/python/cpython/pull/23838 ___ Python tracker <https://bugs.python.org/issue40

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Erlend Egeberg Aasland added the comment: Steve Dower: It should be fairly easy to add a CI check that validates the source checksums, so you don't need to do it manually every time. I can throw up a simple GH workflow for the sqlite branch if you want

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Change by Erlend Egeberg Aasland : -- nosy: +ned.deily, steve.dower ___ Python tracker <https://bugs.python.org/issue42584> ___ ___ Python-bugs-list mailin

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Change by Erlend Egeberg Aasland : -- pull_requests: +22541 pull_request: https://github.com/python/cpython/pull/23675 ___ Python tracker <https://bugs.python.org/issue42

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22540 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23674 ___ Python tracker <https://bugs.python.org/issu

[issue42584] Upgrade macOS and Windows installers to use SQLite 3.34.0

New submission from Erlend Egeberg Aasland : SQLite 3.34.0 was released 2020-12-01: https://www.sqlite.org/releaselog/3_34_0.html Compiles fine on master, and tests are completing without error. -- components: Library (Lib) messages: 382625 nosy: erlendaasland priority: normal

[issue38780] SysLogHandler crash atexit

Erlend Egeberg Aasland added the comment: Thanks, Jason! I won't have time to look at this until perhaps next week, however, feel free to add tests as you like. N.B.: My patches might not be the best solution to the problem; it was just something I threw up based on my initial understanding

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22319 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23428 ___ Python tracker <https://bugs.python.org/issu

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Change by Erlend Egeberg Aasland : -- pull_requests: +22285 pull_request: https://github.com/python/cpython/pull/23393 ___ Python tracker <https://bugs.python.org/issue1635

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Erlend Egeberg Aasland added the comment: For sqlite3, see bpo-42064 -- ___ Python tracker <https://bugs.python.org/issue1635741> ___ ___ Python-bugs-list mailin

[issue42393] Raise overflow errors iso. deprecation warnings in socket.htons and socket.ntohs (was deprecated in 3.7)

New submission from Erlend Egeberg Aasland : Silent truncation was deprecated in Python 3.7 in bpo-28332 with commit 6a7d3480fc8fe4c60538547d2935083ccff0f631 by Serhiy Storchaka. It should be ok to turn these into OverflowError's starting with Python 3.10. -- components: Library (Lib

[issue42393] Raise overflow errors iso. deprecation warnings in socket.htons and socket.ntohs (was deprecated in 3.7)

Change by Erlend Egeberg Aasland : -- type: -> behavior ___ Python tracker <https://bugs.python.org/issue42393> ___ ___ Python-bugs-list mailing list Un

[issue40956] Use Argument Clinic in sqlite3

Change by Erlend Egeberg Aasland : -- pull_requests: +22232 pull_request: https://github.com/python/cpython/pull/23341 ___ Python tracker <https://bugs.python.org/issue40

[issue41861] Convert sqlite3 to heap types (PEP 384)

Erlend Egeberg Aasland added the comment: Yes, all sqlite3 types are converted to heap types. -- resolution: -> fixed stage: patch review -> resolved status: open -> closed ___ Python tracker <https://bugs.python.or

[issue41861] Convert sqlite3 to heap types (PEP 384)

Change by Erlend Egeberg Aasland : -- pull_requests: +6 pull_request: https://github.com/python/cpython/pull/23337 ___ Python tracker <https://bugs.python.org/issue41

[issue41861] Convert sqlite3 to heap types (PEP 384)

Change by Erlend Egeberg Aasland : -- title: Convert sqlite3 to PEP 384 -> Convert sqlite3 to heap types (PEP 384) type: -> resource usage versions: +Python 3.10 ___ Python tracker <https://bugs.python.org/i

[issue29076] Mac installer shell updater script silently fails if default shell is fish

Change by Erlend Egeberg Aasland : -- pull_requests: +22193 stage: needs patch -> patch review pull_request: https://github.com/python/cpython/pull/23302 ___ Python tracker <https://bugs.python.org/issu

[issue29076] Mac installer shell updater script silently fails if default shell is fish

Erlend Egeberg Aasland added the comment: Sure, I'll create a PR right away. -- ___ Python tracker <https://bugs.python.org/issue29076> ___ ___ Python-bug

[issue38823] Improve stdlib module initialization error handling.

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue38823> ___ ___ Python-bugs-list mailing list Unsub

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Change by Erlend Egeberg Aasland : -- pull_requests: +22082 pull_request: https://github.com/python/cpython/pull/23170 ___ Python tracker <https://bugs.python.org/issue1635

[issue42264] Deprecate or remove sqlite3.OptimizedUnicode

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22075 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23163 ___ Python tracker <https://bugs.python.org/issu

[issue42264] Deprecate or remove sqlite3.OptimizedUnicode

New submission from Erlend Egeberg Aasland : Ref. discussion on GH-23148: https://github.com/python/cpython/pull/23148#discussion_r517536083 OptimizedUnicode was obsoleted and undocumented in Python 3.3 (see commit 86dc732f1f). Suggesting to either deprecate OptimizedUnicode in 3.10

[issue42262] [C API] Add Py_NewRef() function to get a new strong reference to an object

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue42262> ___ ___ Python-bugs-list mailing list Unsub

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Change by Erlend Egeberg Aasland : -- pull_requests: +22060 pull_request: https://github.com/python/cpython/pull/23148 ___ Python tracker <https://bugs.python.org/issue1635

[issue15751] [subinterpreters] Make the PyGILState API compatible with subinterpreters

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue15751> ___ ___ Python-bugs-list mailing list Unsub

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22050 pull_request: https://github.com/python/cpython/pull/23136 ___ Python tracker <https://bugs.python.org/issue40

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: FYI, rebased https://github.com/erlend-aasland/cpython/tree/bpo-42064/all onto master, added Petr Viktorin's _PyType_GetModuleByDef() for use in item 7 (module state). I still run into problems in item 8, but I haven't devoted much time for this yet

[issue42100] Add _PyType_GetModuleByDef

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue42100> ___ ___ Python-bugs-list mailing list Unsub

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22040 pull_request: https://github.com/python/cpython/pull/23124 ___ Python tracker <https://bugs.python.org/issue40

[issue1635741] Py_Finalize() doesn't clear all Python objects at exit

Erlend Egeberg Aasland added the comment: I've converted _sre to multi-phase init in this branch: https://github.com/erlend-aasland/cpython/tree/bpo-1635741/sre I'm waiting for GH-23101 to be merged (or rejected) before I submit the PR. It's a fairly large PR (1 file changed, 259 insertions

[issue40077] Convert static types to heap types: use PyType_FromSpec()

Change by Erlend Egeberg Aasland : -- pull_requests: +22025 pull_request: https://github.com/python/cpython/pull/23108 ___ Python tracker <https://bugs.python.org/issue40

[issue42243] Don't access the module dictionary directly

Erlend Egeberg Aasland added the comment: > Are all these occurrences in the module initialization code? Most of them, but not all. > but rewriting can introduce new bugs. Of course. > You can do this, but it will be treated as pure cosmetic change. It can be > accepted if

[issue42243] Don't access the module dictionary directly

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +22018 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23101 ___ Python tracker <https://bugs.python.org/issu

[issue42243] Don't access the module dictionary directly

New submission from Erlend Egeberg Aasland : Quoting from https://docs.python.org/3.10/c-api/module.html: "It is recommended extensions use other PyModule_* and PyObject_* functions rather than directly manipulate a module’s __dict__." A number of modules still access the module

[issue42242] Backport SQLite trace API v2

Erlend Egeberg Aasland added the comment: > [not sure if this is how you do backporting, a new issue and GH PR for each > supported release tag?] Backports are almost always created as backports of a specific GitHub pull request. The pull request is labelled as, for example, "nee

[issue42241] Backport SQLite trace API v2

Change by Erlend Egeberg Aasland : -- keywords: +patch nosy: +erlendaasland nosy_count: 1.0 -> 2.0 pull_requests: +22016 stage: -> patch review pull_request: https://github.com/python/cpython/pull/19581 ___ Python tracker <https://bugs.p

[issue42242] Backport SQLite trace API v2

Change by Erlend Egeberg Aasland : -- keywords: +patch nosy: +erlendaasland nosy_count: 1.0 -> 2.0 pull_requests: +22017 stage: -> patch review pull_request: https://github.com/python/cpython/pull/19581 ___ Python tracker <https://bugs.p

[issue42229] Fix SQLite warnings on macOS

Erlend Egeberg Aasland added the comment: > Should I close this issue—and PR—then? - Backport from master to these? If this fix is wanted in 3.9 and 3.8, I think the correct way to proceed would be to close GH-23072 and backport GH-19581. Try checking out 3.9 and cherrypicking 7f33

[issue42229] Fix SQLite warnings on macOS

Change by Erlend Egeberg Aasland : -- nosy: +ronaldoussoren ___ Python tracker <https://bugs.python.org/issue42229> ___ ___ Python-bugs-list mailing list Unsub

[issue42229] Fix SQLite warnings on macOS

Erlend Egeberg Aasland added the comment: > Yes, this commit extends his changes to include macOS support (for some > reason that SQLite version check doesn’t work properly on macOS I can't reproduce this on master on macOS 10.15.7 with the SQLite 3.28.0 that's shipped with

[issue42229] Fix SQLite warnings on macOS

Erlend Egeberg Aasland added the comment: Oh, I see now that you're only mentioning 3.9 here. The issue was fixed in master only (that is 3.10 alpha). Perhaps GH-19581 should be backported to 3.9, or even 3.8, but that's not for me to decide. Note: your patch is against 3.10, not 3.9

[issue42229] Fix SQLite warnings on macOS

Erlend Egeberg Aasland added the comment: Look at bpo-24464 regarding the shared cache issue. Apparently I did not remember correctly regarding its status. -- ___ Python tracker <https://bugs.python.org/issue42

[issue42229] Fix SQLite warnings on macOS

Erlend Egeberg Aasland added the comment: FYI, commit 7f331c8 (GH-19581 / bpo-40318) migrates the sqlite3 to the new trace API. There was also a discussion about sqlite3_enable_shared_cache() on bpo, but I think it was closed as "wont-fix", IIRC. -- nosy: +erl

[issue40512] [subinterpreters] Meta issue: per-interpreter GIL

Change by Erlend Egeberg Aasland : -- nosy: +erlendaasland ___ Python tracker <https://bugs.python.org/issue40512> ___ ___ Python-bugs-list mailing list Unsub

[issue40956] Use Argument Clinic in sqlite3

Erlend Egeberg Aasland added the comment: > Thank you erlendaasland for working on GH-22484. … and thank you for taking the time to review this! -- ___ Python tracker <https://bugs.python.org/issu

[issue40956] Use Argument Clinic in sqlite3

Change by Erlend Egeberg Aasland : -- pull_requests: +21976 pull_request: https://github.com/python/cpython/pull/23057 ___ Python tracker <https://bugs.python.org/issue40

[issue42213] Get rid of cyclic GC hack in sqlite3.Connection and sqlite3.Cache

Change by Erlend Egeberg Aasland : -- keywords: +patch pull_requests: +21965 stage: -> patch review pull_request: https://github.com/python/cpython/pull/23046 ___ Python tracker <https://bugs.python.org/issu

[issue42213] Get rid of cyclic GC hack in sqlite3.Connection and sqlite3.Cache

New submission from Erlend Egeberg Aasland : Get rid of the `pysqlite_Cache->decref_factory` hack by implementing proper support for cyclic garbage collection. -- components: Library (Lib) messages: 379998 nosy: berker.peksag, erlendaasland priority: normal severity: normal sta

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: I've narrowed down the traverse problems to Py_VISIT(state->CacheType). Suspecting this has to do with the cyclic references in cache/connection. Preliminary tests with GC tracking `cache->factory` look promising, but this is a part of the

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: Sorted out item 7 (module state). Item 8 is mostly in place, although I get ref. count errors if I add traverse/clear callbacks, so I've temporarily disabled those. Proof-of-concept of multi-phase init can be found here: https://github.com/erlend

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: Connection.__init__() fixed by iterating though tp_bases in order to find the correct base class to load the state from. There should be a better way to do this than to dig through type member internals

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: Work with item 7 is progressing; see WIP branch for details. Current showstopper is how to fetch module state from Connection.__init__ when it has been overloaded (for example in /Lib/sqlite3/test/factory.py). Guessing that Py_TYPE(self) does

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: FYI, WIP branch for this bpo is here: https://github.com/erlend-aasland/cpython/commits/bpo-42064/all This currently includes items 1 though 6. I'll update the WIP branch with a draft of items 7 and 8, hopefully tomorrow

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: No, unless we can add `self` as a private context to a sqlite3 database handle, we can't use that shortcut. However, using a custom context struct works, and it's a minor code change. Proof-of-concept implementation diffstat: 2 files changed, 54

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

Erlend Egeberg Aasland added the comment: Instead of using a custom context struct for sqlite3_create_funcion_v2(), we may just use sqlite3_context_db_handle() to get `self`, and then fetch the state using PyType_GetModuleState(Py_TYPE(self

[issue42064] Convert sqlite3 to multi-phase initialisation (PEP 489)

New submission from Erlend Egeberg Aasland : Porting sqlite3 to multi-phase initialisation is non-trivial and must be broken into smaller bpo's, often with multiple PR's per bpo. As far as I can see, this is the task list: 1) Heap types (PEP 384): bpo-41861 2) Argument Clinic: bpo-40956 3

[issue40775] Fix missing dot in sqlite3 Node type name

New submission from Erlend Egeberg Aasland : Fixed in GH-22417 / bpo-41861 -- resolution: -> out of date stage: patch review -> resolved status: open -> closed ___ Python tracker <https://bugs.python.or

[issue42048] Document Argument Clinic's defining_class converter

New submission from Erlend Egeberg Aasland : The 'defining_class' AC converter is not documented. https://www.python.org/dev/peps/pep-0573/#argument-clinic https://docs.python.org/3.10/howto/clinic.html -- components: Argument Clinic messages: 378696 nosy: erlendaasland, larry

[issue42021] Fix possible ref. leaks in _sqlite3 module initialisation

Change by Erlend Egeberg Aasland : -- resolution: -> fixed stage: patch review -> resolved status: open -> closed ___ Python tracker <https://bugs.python.or

[issue40956] Use Argument Clinic in sqlite3

Erlend Egeberg Aasland added the comment: By the way, what's the preferred way to benchmark performance? -- ___ Python tracker <https://bugs.python.org/issue40

<    1   2   3   4   5   6   7   >