New submission from Erlend E. Aasland :
Quoting Petr Viktorin in PR 27940,
https://github.com/python/cpython/pull/27940#discussion_r703424148:
- db is not set to NULL if init fails.
- This segfaults for me:
import sqlite3
conn = sqlite3.connect(':memory:')
conn.execute(&#x
Erlend E. Aasland added the comment:
Quoting Petr from PR 28133:
Without the ability to tell SQLite to abort on trace failure, I doubt we can
do much better than the current behavior. Getting an exception and having
the data inserted seems quite irregular.
Closing this issue as won&#
Change by Erlend E. Aasland :
--
pull_requests: +26634
pull_request: https://github.com/python/cpython/pull/28209
___
Python tracker
<https://bugs.python.org/issue44
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +26648
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/28227
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
> It *would* be possible to improve the documentation, though.
+1
--
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
I modified your second example slightly:
```
import sqlite3
conn = sqlite3.connect(":memory:")
conn.text_factory=bytes
conn.row_factory = sqlite3.Row
cursor = conn.execute("CREATE TABLE foo (bar)")
numbers = range(4)
cursor.executem
Erlend E. Aasland added the comment:
Note: I ran that with PR 28227.
OTOH: I do agree that there's a lot of pitfalls here, especially in the future.
In the long run, it is probably best to deprecate reinit, and disable it in
Python
Erlend E. Aasland added the comment:
Modifying the loops to also print the values:
first fetch
b'0'
b'1'
second fetch
2
3
--
___
Python tracker
<https://bug
Erlend E. Aasland added the comment:
FYI, I've expanded the reinit tests and added a deprecation warning to the PR.
--
___
Python tracker
<https://bugs.python.org/is
Erlend E. Aasland added the comment:
Great, thanks.
--
___
Python tracker
<https://bugs.python.org/issue45126>
___
___
Python-bugs-list mailing list
Unsub
Erlend E. Aasland added the comment:
> If we deprecate but keep the buggy behavior it as it is, (1) is not needed.
> Less work for both us and the users.
Indeed.
There's still a ref leak I'd like to take care of, though: if the first audit
fails, da
Change by Erlend E. Aasland :
--
pull_requests: +26651
pull_request: https://github.com/python/cpython/pull/28231
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
pull_requests: +26654
pull_request: https://github.com/python/cpython/pull/28234
___
Python tracker
<https://bugs.python.org/issue45
Erlend E. Aasland added the comment:
I'll save the cleanup till Python 3.13 dev is started. I've opened a PR for
fixing the ref leak (should be backported), and a draft PR for deprecating
Connection and Cursor reinitialisation.
--
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
nosy_count: 1.0 -> 2.0
pull_requests: +26657
pull_request: https://github.com/python/cpython/pull/28237
___
Python tracker
<https://bugs.python.org/issu
Change by Erlend E. Aasland :
--
pull_requests: +26658
stage: resolved -> patch review
pull_request: https://github.com/python/cpython/pull/28238
___
Python tracker
<https://bugs.python.org/issu
New submission from Erlend E. Aasland :
For SQLite 3.14.0 and newer, we're using the v2 trace API. This means that the
trace callback receives a pointer to the sqlite3_stmt object. We can use the
sqlite3_stmt pointer to retrieve expanded SQL string.
The following stat
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +26660
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/28240
___
Python tracker
<https://bugs.python.org/issu
Change by Erlend E. Aasland :
--
title: [sqlite3] expand bound values in traced statements if possible ->
[sqlite3] expand bound values in traced statements when possible
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
I did some experiments using the connection object as a "backref" in the
callback context, but it seems that the GC does not play well with such ref
circles; I ended up with a lot of ref leaks (yes, I modified the traverse and
clear slots to
Change by Erlend E. Aasland :
--
pull_requests: +26661
pull_request: https://github.com/python/cpython/pull/28242
___
Python tracker
<https://bugs.python.org/issue42
Erlend E. Aasland added the comment:
Updated title to include sqlite3.Cursor as well, since cursors and connections
are very much entwined.
--
title: [sqlite3] cleanup and harden connection init -> [sqlite3] cleanup and
harden Connection and Cursor __ini
Erlend E. Aasland added the comment:
Serhiy, I believe I've addressed your review remarks on PR 27642. Would you
mind taking a look at it again if you have time?
--
___
Python tracker
<https://bugs.python.org/is
Change by Erlend E. Aasland :
--
pull_requests: +26712
pull_request: https://github.com/python/cpython/pull/28298
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
pull_requests: +26713
pull_request: https://github.com/python/cpython/pull/28299
___
Python tracker
<https://bugs.python.org/issue5
New submission from Erlend E. Aasland :
Use unittest.defaultTestLoader instead of unittest.TestLoader() when possible,
to avoid creating unnecessary many instances.
--
components: Tests
messages: 401674
nosy: erlendaasland, serhiy.storchaka
priority: normal
severity: normal
status
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +26715
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/28301
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
> Also, TestLoader has mutable attribute "errors". I am not sure that it is
> good idea to share it between unrelated tests.
Noted. I'm closing this.
--
resolution: -> rejected
stage: patch review -> reso
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45113>
___
___
Python-bugs-list mailing list
Unsubscribe:
Erlend E. Aasland added the comment:
Thanks, Steve. I'll pull in main and see how the CI fares.
--
___
Python tracker
<https://bugs.python.org/is
Erlend E. Aasland added the comment:
_makeLoader is not deprecated yet, so we might keep this open a little bit more.
--
___
Python tracker
<https://bugs.python.org/issue5
Erlend E. Aasland added the comment:
> It is an internal function. We can remove it without deprecation.
How convenient :)
--
___
Python tracker
<https://bugs.python.org/iss
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45116>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Erlend E. Aasland :
I propose to add wrappers for the SQLite sqlite3_limit() C API. Using this API,
it is possible to query and set limits on a connection basis. This will make it
easier (and faster) to test various corner cases in the test suite without
relying on
Erlend E. Aasland added the comment:
Thanks for reviewing, Pablo!
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Erlend E. Aasland added the comment:
Christian, how about adding an audit event for something like
sqlite3.Connection.setlimit()? My initial thought is: yes.
--
nosy: +christian.heimes
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +26864
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/28463
___
Python tracker
<https://bugs.python.org/issu
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue43760>
___
___
Python-bugs-list mailing list
Unsubscribe:
Change by Erlend E. Aasland :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Change by Erlend E. Aasland :
--
pull_requests: +26886
pull_request: https://github.com/python/cpython/pull/28490
___
Python tracker
<https://bugs.python.org/issue44
Erlend E. Aasland added the comment:
Now that pysqlite_statement_reset() is only used in cursor.c, I suggest to move
it to cursor.c and make it a static function.
--
___
Python tracker
<https://bugs.python.org/issue44
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45255>
___
___
Python-bugs-list mailing list
Unsubscribe:
Erlend E. Aasland added the comment:
> Could it instead throw a FileNotFoundError if the db simply doesn't exist at
> the specified path?
We pass the database path pretty much straight to sqlite3_open_v2(), via the
os.PathLike interface, so any errors we get come straight from SQ
Change by Erlend E. Aasland :
--
status: open -> pending
___
Python tracker
<https://bugs.python.org/issue45255>
___
___
Python-bugs-list mailing list
Un
Change by Erlend E. Aasland :
--
pull_requests: +26906
pull_request: https://github.com/python/cpython/pull/28509
___
Python tracker
<https://bugs.python.org/issue45
Erlend E. Aasland added the comment:
> Now that pysqlite_statement_reset() is only used in cursor.c, I suggest to
> move it to cursor.c and make it a static function.
I'll wait until PR 25984 is merged before opening a PR for relocating
pysqlite_stat
Erlend E. Aasland added the comment:
Ouch, that's quite a regression! Thanks for the heads up! I'll have a look at
it right away.
--
___
Python tracker
<https://bugs.python.o
Erlend E. Aasland added the comment:
I'm unable to reproduce this regression on my machine (macOS, debug build, no
optimisations). Are you able to reproduce, Ken?
--
___
Python tracker
<https://bugs.python.org/is
Erlend E. Aasland added the comment:
> I'm unable to reproduce this regression on my machine (macOS, debug build, no
> optimisations) [...]
Correction: I _am_ able to reproduce this.
--
___
Python tracker
<https://bugs.python.
Erlend E. Aasland added the comment:
Explicitly resetting statements when we're done with them removes the
performance regression; SQLite works more efficient when we keep the number of
non-reset statements low.
--
___
Python tracker
&
Erlend E. Aasland added the comment:
I'll revert PR 27844 for now (except the tests).
Since SQLite works better when we keep the number of non-reset statements to a
minimum, we need to ensure that we reset statements when we're done with them
(sqlite3_step() returns SQLITE_DONE o
Change by Erlend E. Aasland :
--
pull_requests: +26956
pull_request: https://github.com/python/cpython/pull/28574
___
Python tracker
<https://bugs.python.org/issue44
Change by Erlend E. Aasland :
--
resolution: wont fix -> fixed
___
Python tracker
<https://bugs.python.org/issue45089>
___
___
Python-bugs-list mailing list
Un
Erlend E. Aasland added the comment:
Replacing the timestamp converter with datetime.datetime.fromisoformat sounds
good to me.
--
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue29410>
___
___
Python-bugs-list mailing list
Unsubscribe:
Erlend E. Aasland added the comment:
> [...] if sqlite3 starts returning aware datetimes, existing code might break.
True.
> [...] perhaps this could be fixed in conjunction with changing sqlite3's API
> to allow per-database converters and adapters
Another option could be to
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45374>
___
___
Python-bugs-list mailing list
Unsubscribe:
Erlend E. Aasland added the comment:
> I'd be interested in working on this myself, if you think it's something that
> a new CPython contributor could handle.
Please, go ahead :)
However, I think this should be discussed on Discourse first (open a topic in
the Core Deve
Erlend E. Aasland added the comment:
See also:
- bpo-19065: sqlite3 timestamp adapter chokes on timezones
- bpo-26651 Deprecate register_adapter() and register_converter() in sqlite3
(Adding Berker to nosy list.)
--
nosy: +berker.peksag
Erlend E. Aasland added the comment:
Yes, deprecating the preconfigured adapters and converters, but keeping the
ability to register adapters/converters sounds like a very good idea to me. As
you point out, Marc-Andre, they are application specific. Also, implementing
converters and
Change by Erlend E. Aasland :
--
versions: +Python 3.11 -Python 3.6
___
Python tracker
<https://bugs.python.org/issue26651>
___
___
Python-bugs-list mailin
Change by Erlend E. Aasland :
--
pull_requests: +27123
pull_request: https://github.com/python/cpython/pull/28790
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
pull_requests: +27128
pull_request: https://github.com/python/cpython/pull/28809
___
Python tracker
<https://bugs.python.org/issue16
Erlend E. Aasland added the comment:
FYI: We've been using sqlite3_close_v2(), when available, since 2017 (see
86a670543ff97d52fd9b8ca0477f8b6d27ee946d), but we now call it with GIL held.
I'm leaning towards closing this as out-of-date. There is no reproducer, there
has been no mo
Erlend E. Aasland added the comment:
>From https://sqlite.org/c3ref/close.html:
If sqlite3_close_v2() is called with unfinalized prepared statements,
unclosed BLOB handlers, and/or unfinished sqlite3_backups, it returns
SQLITE_OK regardless, but instead of deallocating the datab
New submission from Erlend E. Aasland :
While working on bpo-45126 / GH-28227 (and while working on the AC conversion,
bpo-40956), I've been slightly frustrated on the implementation of sqlite3
"isolation level". The code is messy, and we've got two connection members tha
Erlend E. Aasland added the comment:
+1!
See also bpo-43502
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45490>
___
___
Python-bug
Erlend E. Aasland added the comment:
Previous discussion on Discourse:
https://discuss.python.org/t/what-to-do-with-unsafe-macros/7771
--
___
Python tracker
<https://bugs.python.org/issue45
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +27325
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/29053
___
Python tracker
<https://bugs.python.org/issu
Change by Erlend E. Aasland :
--
pull_requests: +27326
pull_request: https://github.com/python/cpython/pull/29054
___
Python tracker
<https://bugs.python.org/issue36
Erlend E. Aasland added the comment:
If you are building CPython against a specific version of SQLite (for example
one you've compiled yourself), make sure that you adjust _both_ CFLAGS and
LDFLAGS.
Note: since bpo-40810 / GH-24106, SQLite 3.7.15 or newer is required.
Closing this as
Change by Erlend E. Aasland :
--
resolution: -> fixed
stage: patch review -> resolved
status: open -> closed
___
Python tracker
<https://bugs.python.or
Erlend E. Aasland added the comment:
I'm closing this for now. As I see it, there is little added value, and it
would be hard to make a good API for all invariants. If this idea surfaces
again, we can always reopen this issue.
Thanks for you interest in improving CP
Change by Erlend E. Aasland :
--
pull_requests: +27341
pull_request: https://github.com/python/cpython/pull/29073
___
Python tracker
<https://bugs.python.org/issue42
Erlend E. Aasland added the comment:
It seems to me that this issue can be closed as fixed. Do you have further PR's
for this, Petr?
--
nosy: +erlendaasland
status: open -> pending
___
Python tracker
<https://bugs.python.org
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45459>
___
___
Python-bugs-list mailing list
Unsubscribe:
Erlend E. Aasland added the comment:
FYI, see bpo-43916 regarding the introduction and rationale of introducing
Py_TPFLAGS_DISALLOW_INSTANTIATION.
Slightly related: bpo-43908
--
___
Python tracker
<https://bugs.python.org/issue45
Erlend E. Aasland added the comment:
>> Can I do something for this pending PEP?
>
> Ask Victor, he should know more. But as far as I know, no one started on it
> yet.
Quoting PEP 630 (active PEP):
Whenever this PEP mentions extension modules, the advice also applies to
b
Erlend E. Aasland added the comment:
> I no longer think that we should define Py_BUILD_CORE_MODULE
> unconditionally. Instead I propose to move the defines into each C module.
+1. Explicit is nice.
--
nosy: +erlendaasland
___
Python t
Change by Erlend E. Aasland :
--
pull_requests: +27435
pull_request: https://github.com/python/cpython/pull/29160
___
Python tracker
<https://bugs.python.org/issue45
Erlend E. Aasland added the comment:
> My understanding is that this entire class of code changes has been put on
> hold pending Steering Council approval.
Can you please point me to the official SC statement regarding this? I cannot
find it.
> It represents a great deal of code
New submission from Erlend E. Aasland :
Currently, we call _pysqlite_seterror() if sqlite3_open_v2() returns !=
SQLITE_OK. However, if a memory failure occurs during sqlite3_open_v2(), the
database handle is explicitly set to NULL. This _may_ cause
_pysqlite_seterror() to segfault, since we
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +27444
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/29171
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
New failure today:
https://github.com/python/cpython/runs/3979900415?check_suite_focus=true
test_sendfile_close_peer_in_the_middle_of_receiving
(test.test_asyncio.test_sendfile.SelectEventLoopTests) ... ok
Erlend E. Aasland added the comment:
Thanks, Petr. Yes, that is correct. PEP 630 (I assume you meant 630, not 640)
still describes the _rationale_ very well. I guess what's needed is a Standards
Track PEP based on PEP 630 (IIUC).
--
___
P
Erlend E. Aasland added the comment:
Petr:
> Hmm, the best I can find is
> https://github.com/python/steering-council/blob/main/updates/2021-02-steering-council-update.md#february-08
Perfect, thanks!
Christian: Thanks for your input. If this was a discussion on Discourse, I'd
Erlend E. Aasland added the comment:
Oh, I found PEP 3121 (Extension Module Initialization and Finalization). It is
a Standards Track PEP and it is accepted.
The abstract is pretty short. Let me just repost it here, for convenience:
Extension module initialization currently has a few
New submission from Erlend E. Aasland :
Some DB-API 2.0 attributes are undocumented:
- apilevel
- threadsafety
- paramstyle
These attributes should be documented.
See also:
- bpo-8196
-
https://discuss.python.org/t/is-sqlite3-threadsafety-the-same-thing-as-sqlite3-threadsafe-from-the-c
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +27482
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/29219
___
Python tracker
<https://bugs.python.org/issu
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue37095>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Erlend E. Aasland :
Currently, the sqlite3 module docstring is the old pysqlite copyright comment
in the start of Lib/sqlite3/__init__.py. This is not very helpful to the
general user. Suggesting to add a more helpful docstring, a la the 're' or
'array'
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +27488
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/29224
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
A side-effect of keeping tests inside the package, is that they appear in the
package contents module docstring, which may be slightly confusing to new users.
>>> import sqlite3
>>> help(sqlite3)
Help on package sqlite3:
NAME
sq
Change by Erlend E. Aasland :
--
nosy: +serhiy.storchaka
___
Python tracker
<https://bugs.python.org/issue45613>
___
___
Python-bugs-list mailing list
Unsub
Change by Erlend E. Aasland :
--
nosy: +paul.moore
___
Python tracker
<https://bugs.python.org/issue45613>
___
___
Python-bugs-list mailing list
Unsubscribe:
New submission from Erlend E. Aasland :
Currently, the sqlite3 DB-API 2.0 attribute 'threadsafety' is hard-coded to 1,
meaning "threads may share the module, but not connections". This is not always
true, since it depends on the default SQLite threaded mode, selected at
c
Erlend E. Aasland added the comment:
FYI, it is also possible to change the threaded mode using sqlite3_config(),
however that API is not exposed to the sqlite3 module, and there is no plans
for doing so in the near future :)
--
___
Python
Change by Erlend E. Aasland :
--
keywords: +patch
pull_requests: +27491
stage: -> patch review
pull_request: https://github.com/python/cpython/pull/29227
___
Python tracker
<https://bugs.python.org/issu
Erlend E. Aasland added the comment:
BTW, I've verified that the overhead of adding this query is negligible.
$ python -m pyperf compare_to --min-speed 1 main.json patched.json
Benchmark hidden because not significant (1): sqlite_synth
git switch -
Switched to branch 'main'
Change by Erlend E. Aasland :
--
nosy: +erlendaasland
___
Python tracker
<https://bugs.python.org/issue45614>
___
___
Python-bugs-list mailing list
Unsubscribe:
601 - 700 of 829 matches
Mail list logo
