Daniel Fett

Yes, can confirm, no significant CPU load caused by either emacs or elpy process.

Will run the debug script as soon as the next lockup happens (restarted elpy just now). NB: strace should show new output when I scroll through a file (it should...

Emacs: (utf-8-unix . utf-8-unix) Python: '\n'

Well, I tried the debug script above, but it does not create any output when elpy has crashed and I have to abort with C-g.

This is what it looked like: elpy-rpc--call duration 0.0004696846008300781

Different "projects" yield different pids, yes. It looks like the longer the messages are, the earlier the crash happens (smells like a buffer running full). The script above does not...

I added the following code to read_json: ``` line = "" with open('/tmp/elpy-debug-python-server.log', 'a') as f: f.write(str(os.getpid())) f.write(" ") while True: byte = self.stdin.read(1) f.write(byte) f.flush() line += byte if...

Oh, but it also never stops with just the PID written to the file, meaning that elpy does not wait for new messages, but is occupied itself.

Yes, that came to me just right now. With flush, the output ends after the pid. So the server IS waiting.

Yes. Tried four times, crashed between message 518 and 1350.