Call Application.startup in GunicornWebWorker

[f0t0n]

What do these changes do?

• Introduced the same behavior as in web.run_app() to call Application.startup() along with the request handler.
• Extended test case to check worker.wsgi.startup has been called once.

Are there changes in behavior for the user?

None

Related issue number

#1092

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes

[asvetlov]

Please upgrade to master.

appveyor blames:

=================================== ERRORS ====================================
____________________ ERROR collecting tests/test_worker.py ____________________
Using @pytest.skip outside a test (e.g. as a test function decorator) is not allowed. Use @pytest.mark.skip or @pytest.mark.skipif instead.

But the master has no pytest.skip anymore in test_worker.py

[f0t0n]

It did at #1103 as well. The branch is rebased and is up to date.

[codecov-io]

Current coverage is 98.06% (diff: 100%)

Merging #1105 into master will increase coverage by <.01%

@@             master      #1105   diff @@
==========================================
  Files            28         28          
  Lines          6364       6365     +1   
  Methods           0          0          
  Messages          0          0          
  Branches       1070       1070          
==========================================
+ Hits           6241       6242     +1   
  Misses           64         64          
  Partials         59         59          

Powered by Codecov. Last update ec83698...d90bbbe

[asvetlov]

I suspect parallel execution by gather is not desired.
At least it works different that web.run_app().
Better to wait for self.wsgi.startup() and than execute self._runner for opening server-side socket etc.

[f0t0n]

I'm not sure I get you right. Do you mean to just schedule self.wsgi.startup() with ensure_future() and keep the rest as it was done before?

ensure_future(self.wsgi.startup(), loop=self.loop)
self._runner = ensure_future(self._run(), loop=self.loop)

try:
    self.loop.run_until_complete(self._runner)
finally:
    self.loop.close()

[f0t0n]

Fixed as you suggested.

[asvetlov]

Every future worth to be awaited :)

self.loop.run_until_complete(self.wsgi.startup())
self._runner = ensure_future(self._run(), loop=self.loop)

try:
    self.loop.run_until_complete(self._runner)
finally:
    self.loop.close()

Initialize app on first step and run it on the second.
Wait for end of finalization process before starting the app.

[f0t0n]

Ah, I already thought to yield from self.wsgi.startup() it in self._run(), let's keep your version.

[asvetlov]

Thanks!

[lock]

This thread has been automatically locked since there has not been
any recent activity after it was closed. Please open a new issue for
related bugs.

If you feel like there's important points made in this discussion,
please include those exceprts into that new issue.