[sqlite3] _pysqlite_connection_begin() optimisations

[erlend-aasland]

BPO	44042
Nosy	@berkerpeksag, @serhiy-storchaka, @erlend-aasland
PRs	bpo-44042: sqlite3 begin transaction optimisations #25908

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2021-06-03.21:24:34.944>
created_at = <Date 2021-05-04.23:41:24.768>
labels = ['3.11', 'performance']
title = '[sqlite3]  _pysqlite_connection_begin() optimisations'
updated_at = <Date 2021-06-03.21:24:34.943>
user = 'https://github.com/erlend-aasland'

bugs.python.org fields:

activity = <Date 2021-06-03.21:24:34.943>
actor = 'pablogsal'
assignee = 'none'
closed = True
closed_date = <Date 2021-06-03.21:24:34.944>
closer = 'pablogsal'
components = []
creation = <Date 2021-05-04.23:41:24.768>
creator = 'erlendaasland'
dependencies = []
files = []
hgrepos = []
issue_num = 44042
keywords = ['patch']
message_count = 2.0
messages = ['392967', '395051']
nosy_count = 3.0
nosy_names = ['berker.peksag', 'serhiy.storchaka', 'erlendaasland']
pr_nums = ['25908']
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'performance'
url = 'https://bugs.python.org/issue44042'
versions = ['Python 3.11']

[erlend-aasland]

The following optimisations can be applied to _pysqlite_connection_begin():

1. Return an int instead of a PyObject pointer

   Per now, we do Py_RETURN_NONE and Py_DECREF(result) if _pysqlite_connection_begin() was successful (normally the case). There's no reason to do this. Let's just it the C way: return -1 on error and 0 if things are ok.

2. Defer error checking till post sqlite3_finalize()

   Any error code returned by sqlite3_step() will also be returned by sqlite3_finalize() for the same statement. From the SQLite docs:
   "If the most recent evaluation of statement S failed, then sqlite3_finalize(S) returns the appropriate error code or extended error code."

3. Move _pysqlite_connection_begin() to Modules/_sqlite/cursor.c

   The single use is in _pysqlite_query_execute() in cursor.c. Moving it makes it possible for the compiler to apply more optimisations. At least so I've heard :)
   (As a side effect, the namespace will be cleaner.)

[pablogsal]

New changeset 3446516 by Erlend Egeberg Aasland in branch 'main':
bpo-44042: Optimize sqlite3 begin transaction (GH-25908)
3446516