Use check_same_thread=False by default for history sqlite db (#13886)...
Use check_same_thread=False by default for history sqlite db (#13886)
I had run into the following exception while trying to use IPython in a
thread:
```
Traceback (most recent call last):
File "/usr/local/lib/python3.11/dist-packages/IPython/core/interactiveshell.py", line 3745, in atexit_operations
self._atexit_once()
File "/usr/local/lib/python3.11/dist-packages/IPython/core/interactiveshell.py", line 3728, in _atexit_once
self.history_manager.end_session()
File "/usr/local/lib/python3.11/dist-packages/IPython/core/history.py", line 576, in end_session
self.writeout_cache()
File "/usr/local/lib/python3.11/dist-packages/decorator.py", line 232, in fun
return caller(func, *(extras + args), **kw)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/IPython/core/history.py", line 60, in only_when_enabled
return f(self, *a, **kw)
^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.11/dist-packages/IPython/core/history.py", line 831, in writeout_cache
self._writeout_input_cache(conn)
File "/usr/local/lib/python3.11/dist-packages/IPython/core/history.py", line 812, in _writeout_input_cache
with conn:
sqlite3.ProgrammingError: SQLite objects created in a thread can only be used in that same thread. The object was created in thread
id
139673788811008 and this is thread id 139673823184704.
```
And discovered that an issue (#680) has been open for it since 2011.
Back in
[2012](
https://github.com/ipython/ipython/issues/680#issuecomment-3444922),
it seems like the only reason not to fix it was that the parameter
`check_same_thread` was not documented, but now [it
is](
https://docs.python.org/3.11/library/sqlite3.html#sqlite3.connect),
and it has been at least since
[3.6](
https://docs.python.org/3.6/library/sqlite3.html#sqlite3.connect).
Note that according to the docs:
> check_same_thread
([bool](
https://docs.python.org/3/library/functions.html#bool)) – If
True (default), only the creating thread may use the connection. If
False, the connection may be shared across multiple threads; if so,
write operations should be serialized by the user to avoid data
corruption.
But I don't think this is an issue here. The operations should be
synchronized by the user only *on the same connection object*, and if
I'm not mistaken, if two instances of IPython were started in separate
threads, they'd each have their own history manager with its own
connection. The exception above (and the one in the related issue) is
raised from `atexit`, when the main thread is running finalizations
registered by other threads. At that point, the operations are serial
anyway because they are executed in a single thread.