RuntimeError: This event loop is already running #326
Labels
Comments
|
Tried the following patch: diff --git i/neovim/msgpack_rpc/event_loop/asyncio.py w/neovim/msgpack_rpc/event_loop/asyncio.py
index 656f552..5289957 100644
--- i/neovim/msgpack_rpc/event_loop/asyncio.py
+++ w/neovim/msgpack_rpc/event_loop/asyncio.py
@@ -106,12 +106,26 @@ def _start_reading(self):
def _send(self, data):
self._transport.write(data)
+ def is_running(self):
+ return self._loop.is_running() and not self._loop._stopping
+
def _run(self):
while self._queued_data:
self._on_data(self._queued_data.popleft())
diff --git i/neovim/msgpack_rpc/session.py w/neovim/msgpack_rpc/session.py
index b300f8d..1f93c6f 100644
--- i/neovim/msgpack_rpc/session.py
+++ w/neovim/msgpack_rpc/session.py
@@ -87,7 +87,7 @@ def request(self, method, *args, **kwargs):
raise ValueError("request got unsupported keyword argument(s): {}"
.format(', '.join(kwargs.keys())))
- if self._is_running:
+ if self._is_running and self.loop.is_running():
v = self._yielding_request(method, args)
else:
v = self._blocking_request(method, args)But that does not fix it. I guess it might miss events, and I've also seen this then: |
|
Hmm, it looks like nvim methods are invoked directly from a separate thread, that is definietly not safe. |
|
Related? #324 / tweekmonster/deoplete-clang2#30 |
This was referenced May 16, 2018
|
@bfredl |
|
With deoplete-jedi it currently looks like this: |
I don't think I can. Deoplete doesn't use threading anymore. |
|
deoplete-jedi does though. It basically can be triggered by using deoplete's logger from the thread. deoplete's logger has a filter to call into Neovim via |
|
I wonder if deoplete-jedi should also switch to using asyncio instead of threads, but that might not be trivial. /cc @Shougo |
|
As for deoplete-jedi I will change it to get the exceptions from the threads into the caller, and then logging should happen from there. |
Hm. If possible. |
|
Can we close this? Can someone suggest a small documentation update? |
|
What would be a documentation update? |
|
This is what I've done for deoplete-jedi: deoplete-plugins/deoplete-jedi#178 - it stores exceptions in the threads, and collects them for non-alive threads then. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I am seeing a
RuntimeError: This event loop is already runningwhen using`log.exception from deoplete-jedi's server.
This appears to be caused by DeopleteLogFilter.filter using
self.vim.call('deoplete#util#print_error', record.getMessage(), record.name).Full traceback (the code in deoplete-jedi (from the first exception) is new ):
Another instance:
Code in cpython: https://github.com/python/cpython/blob/20d68dfcc07bd389ce2ea7b0773c44d97ebeb68d/Lib/asyncio/base_events.py#L507-L508
It appears to be a race, since
is_running()only looks at the thread_id, andnot
_stopping(which is whatstop()sets).Events:
So neovim stops the loop ("stopping"), but then calls
self._loop.run_foreveragain, since the internalis_runningis false, butrun_foreverwill then raise the RuntimeError.I've tried using
self.loop._loop.is_runningdirectly/alone, but then the greenlet parent might be None:The text was updated successfully, but these errors were encountered: