1. 23 Aug, 2019 10 commits
    • Kirill Smelkov's avatar
      golang: tests: Factor out retrieving len(ch._recvq) and len(ch._sendq) · 352628b5
      Kirill Smelkov authored
      Chan implementation is going to be moved into C, and this way direct
      access to chan internals won't be possible. C implementation will export
      dedicated functions for chan tests which will be used.
      
      Prepare for that and factor out retrieving len of chan recv/send queues
      into separate functions, which will be adapted at the time of C port.
      352628b5
    • Kirill Smelkov's avatar
      golang: Don't verify ._recvq and ._sendq of nil channel · d98e42e3
      Kirill Smelkov authored
      Nil channel will soon be changed to be represented by underlying NULL C
      pointer and there won't be channel object for nil and correspondingly
      recv/send queues won't be there. -> Prepare for that and don't check for
      recvq/sendq on nil channel.
      d98e42e3
    • Kirill Smelkov's avatar
      golang: Remove outdated TODO · 049ba6ff
      Kirill Smelkov authored
      We provide gpython since 32a21d5b (gpython: Python interpreter with
      support for lightweight threads), and golang module, since the
      beginning, automatically uses gevent if it was installed via monkey
      patching.
      049ba6ff
    • Kirill Smelkov's avatar
      time: Test for now · 7f2362dd
      Kirill Smelkov authored
      This currently pass trivially (since time.now = stdtime.now), but will
      be useful once we get time.now implementation into our hands and move it into Cython.
      7f2362dd
    • Kirill Smelkov's avatar
      *: Use golang.time universally · c3bd2c50
      Kirill Smelkov authored
      After 9c260fde (time: New package that mirrors Go's time) we have
      golang.time.now and golang.time.sleep and it makes it a more
      self-dependent system if timing facility is used through golang.time
      instead of outside std time module.
      
      For now this is only a "cleanness" change, but will become important once
      we start adding pyx-level nogil API to pygolang - there it will be
      important to use golang.time.* for correctness.
      c3bd2c50
    • Kirill Smelkov's avatar
      sync: threading.Event -> chan · 78d85cdc
      Kirill Smelkov authored
      Starting from e6bea2cf (sync: New package that mirrors Go's sync) there
      is sync.WaitGroup and it was using threading.Event to signal that
      waitgroup's counter dropped to zero. threading.Event is implemented in
      Python's stdlib via threading.Condition, which in turn uses
      threading.Lock and list of waiters to implement its functionality.
      Which in turn is very similar to what golang.chan does internally with
      semaphore and e.g. recv queue.
      
      I noticed this while debugging sync test deadlock (see previous patch)
      and suspecting bug in threading.Event for a moment. While it turned
      there is no bug in threading.Event, it is better to rely on the common
      functionality for similar tasks, and pygolang's channel perfectly
      matches here the need to signal an event. By using our own code instead
      of stdlib's threading we are likely becoming less bug-prone.
      
      This also brings a small speedup:
      
          (on i5@1.80GHz)
          name             old time/op  new time/op  delta
          workgroup_empty   239µs ± 1%   220µs ± 1%  -7.62%  (p=0.000 n=10+9)
          workgroup_raise   275µs ± 2%   264µs ± 2%  -3.85%  (p=0.000 n=10+10)
      78d85cdc
    • Kirill Smelkov's avatar
      sync.WorkGroup: Fix deadlock thinko in tests · b8b042c5
      Kirill Smelkov authored
      sync.WorkGroup test was doing ctx.done() wait from under test mutex,
      something like
      
          def _(ctx, i):
              with mu:
                  ...
                  if i == 0:
                      raise RuntimeError()    # to cause ctx cancel
                  ctx.done().recv()           # i=1 -> wait till ctx is canceled
      
      but it can be a deadlock if T(i=1) runs first and enters
      ctx.done().recv() before T(i=0) is run - then T(i=0) will block forever
      waiting to lock mu.
      
      This failure was not seen so far, probably because the time to go a new
      thread/goroutine is relatively high. However one of upcoming patches,
      where go is made faster, revealed this problem and, without the fix,
      sync.WorkGroup test was regularly deadlocking.
      
      The problem was there from sync.WorkGroup beginning - from
      9ee7ba91 (sync += WorkGroup).
      
      Fix the deadlock by waiting for ctx.done() outside of mu.
      b8b042c5
    • Kirill Smelkov's avatar
      golang: gimport works on Py3 · 086d4296
      Kirill Smelkov authored
      9c61f254 (pygopath: Initial draft) has a "TODO py3 support", probably
      because https://stackoverflow.com/a/67692 shows different codes for
      Python2 and Python3.
      
      However today we have tox coverage for all Python2.7, Python3.6 and
      Python3.7 and everything works including gimport tests.
      
      -> Remove the TODO.
      086d4296
    • Kirill Smelkov's avatar
      golang/_gopath_test: Redo $GOPATH set/restore via fixture · 7d08f078
      Kirill Smelkov authored
      Currently gimport tests depend on module-level setup_module /
      teardown_module, which require for gimport tests to be in separate file.
      
      Redo the test not to depend on global module-level state and
      setup/teardown. This allows e.g. to merge gimport tests into
      golang_test.py if/when we want/need to.
      7d08f078
    • Kirill Smelkov's avatar
      golang: Test len(nilchan) and repr(nilchan) · 3e5b5f01
      Kirill Smelkov authored
      Should be in 2aad64bb (golang: Add support for nil channel).
      3e5b5f01
  2. 20 Jul, 2019 1 commit
    • Kirill Smelkov's avatar
      *_test: Verify panic argument · 773d8fb2
      Kirill Smelkov authored
      In tests in places where the code checks that something panics, verify
      not only that _PanicError is raised, but also what was the argument
      passed to panic.
      
      This makes sure that tested code panics in expected place, not just
      panics "somewhere".
      
      To keep signal/noise ratio high introduce `panics` which is similar to
      `pytest.raises` and asserts that wrapped code panics with expected
      argument.
      773d8fb2
  3. 08 Jul, 2019 1 commit
  4. 26 Jun, 2019 4 commits
    • Kirill Smelkov's avatar
      sync.WorkGroup: Propagate all exception types, not only those derived from Exception · 79aab7df
      Kirill Smelkov authored
      A problem was hit with pytest.fail with raises Failed exception not
      being propagated to .wait. As it turned out it was not propagated
      because pytest's Failed derives from BaseException, not Exception, and
      we were catching only Exception and its children.
      
      Rework the code to propagate all exception types from workers.
      
      Performance change is with noise (it is either a bit faster for one set
      of runs, or a bit slower for another set of runs).
      79aab7df
    • Kirill Smelkov's avatar
      sync.Workgroup: Don't use @func at runtime · 94c6160b
      Kirill Smelkov authored
      It is true for any decorator, that it makes things faster if
      def+decorator is used globally instead of at runtime, but for @func it
      is especially true since @func, using decorator.decorate, has relatively
      high overhead if it is used not only once at program startup, but
      instead every time something is run.
      
      In particular moving @func out of WorkGroup.go() make things
      considerably faster:
      
      	name             old time/op  new time/op  delta
      	workgroup_empty   195µs ± 0%   113µs ± 1%  -41.91%  (p=0.008 n=5+5)
      	workgroup_raise   221µs ± 1%   137µs ± 1%  -38.29%  (p=0.008 n=5+5)
      
      See
      
      	https://lab.nexedi.com/kirr/misc/raw/009c4fee/pygolang/prof_workgroup_empty.svg
      
      for bench_workgroup_empty profile, where it is seen that
      decorator.decorate was using ~ half of the whole WorkGroup.go() time.
      94c6160b
    • Kirill Smelkov's avatar
      golang: Add benchmarks for chan, select, @func and defer · 3c55ca59
      Kirill Smelkov authored
      Start adding benchmarks to pygolang.
      
      Measure how much sync chan send/recv take and how much select takes for
      synchronous channels. Also measure how much overhead @func adds at both
      def and call times, and the overhead of defer compared to try/finally.
      For std (non-gevent'ed) Python2.7 we are currently at:
      
      	name         time/op
      	go           91.6µs ± 1%
      	chan         13.7µs ± 3%
      	select       30.1µs ± 4%
      	def          55.0ns ± 0%
      	func_def     43.6µs ± 0%
      	call         63.0ns ± 0%
      	func_call    1.06µs ± 0%
      	try_finally   136ns ± 1%
      	defer        2.33µs ± 0%
      3c55ca59
    • Kirill Smelkov's avatar
      golang: test_method -> test_func · c3ea8f51
      Kirill Smelkov authored
      @method(cls) was deprecated and removed in favour of @func(cls) in
      942ee900 (golang: Deprecate @method(cls) in favour of @func(cls)) and
      262f8986 (golang: Kill @method). The test name was overlooked.
      c3ea8f51
  5. 21 Jun, 2019 1 commit
    • Kirill Smelkov's avatar
      golang: Fix `@func(cls) def name` not to override `name` in calling context · 924a808c
      Kirill Smelkov authored
      With @func being a decorator, the following
      
      	@func(cls)
      	def name():
      		...
      
      is always processed by python as
      
      	name = func(cls)(def name(): ...)
      
      Before this patch it was leading to name being overridden with None:
      
      	def f():
      		print 'hello'
      
      	class C:
      		pass
      
      	@func(C)
      	def f(c):
      		print 'C.f', c
      
      	f()
      	Traceback (most recent call last):
      	  File "<console>", line 1, in <module>
      	TypeError: 'NoneType' object is not callable
      
      We can fix it by returning from `func(cls)(def name(): ...)` the
      original `name` object from the calling context.
      
      Unfortunately if `name` was not previously set I did not find a way(*) to
      avoid polluting the calling namespace where it is set to what @func(cls)
      returns (None) by the hardcoded way how python processes decorators:
      
      	In [2]: c = """
      	   ...: @fff
      	   ...: def ccc():
      	   ...:     return 1
      	   ...: """
      
      	In [3]: cc = compile(c, "file", "exec")
      
      	In [4]: dis(cc)
      	  2           0 LOAD_NAME                0 (fff)
      	              3 LOAD_CONST               0 (<code object ccc at 0x7fafe58d0130, file "file", line 2>)
      	              6 MAKE_FUNCTION            0
      	              9 CALL_FUNCTION            1
      	             12 STORE_NAME               1 (ccc)	<-- NOTE means: ccc = what fff() call returns
      	             15 LOAD_CONST               1 (None)
      	             18 RETURN_VALUE
      
      At least with no overriding taking place the situation is better now.
      
      NOTE: it is only @func(cls) which potentially pollutes calling
      namespace. Just @func (without class) is always clean because by
      definition it works as a regular decorator.
      
      (*) there is a very low-level and potentially fragile way to disable
      STORE_NAME after CALL_FUNCTION by dynamically patching caller's bytecode
      at runtime and replacing STORE_NAME with POP_TOP + NOP...
      924a808c
  6. 24 May, 2019 1 commit
    • Kirill Smelkov's avatar
      Revert "tox -= Python3.6" · 08ec7950
      Kirill Smelkov authored
      This reverts commit 469f21a9.
      
      Even though Debian stopped shipping Python3.6 I'm now building it myself
      to have test coverage for two latest Python releases.
      08ec7950
  7. 16 May, 2019 7 commits
    • Kirill Smelkov's avatar
      pygolang v0.0.2 · d3bfb1bf
      Kirill Smelkov authored
      d3bfb1bf
    • Kirill Smelkov's avatar
      context: Add support for deadlines · b2450310
      Kirill Smelkov authored
      Implement deadlines / timeouts using timers added recently in 9c260fde
      (time: New package that mirrors Go's time).
      b2450310
    • Kirill Smelkov's avatar
      context: tests: Factor out common bits from test_context · 27f91b78
      Kirill Smelkov authored
      In the next patch we'll be adding deadlines support with another test.
      Move common infrastructure that will be used in all context tests to be
      test-module global.
      27f91b78
    • Kirill Smelkov's avatar
      context: Don't hardcode that contexts are always done with canceled error · e5687f2f
      Kirill Smelkov authored
      We'll need to mark contexts that miss their deadline with
      deadlineExceeded error in a follow-up patch.
      
      In this patch: transform _BaseCtx._cancel to accept an error with which
      a context has to be canceled, but for now always use canceled in places
      where _cancel is called.
      e5687f2f
    • Kirill Smelkov's avatar
      context: Note that a context has to be explicitly canceled to release resources · 58ba1765
      Kirill Smelkov authored
      Same story as in Go, but we were lacking documentation notice about this.
      58ba1765
    • Kirill Smelkov's avatar
      time: New package that mirrors Go's time · 9c260fde
      Kirill Smelkov authored
      Add time.Timer, time.Ticker and convenience functions time.tick,
      time.after and time.after_func. These will be used in context to support
      deadlines and timeouts.
      
      While at time topic, also provide sleep and now from golang.time, so
      that there is no need to import both golang.time and stdlib's time in a
      file.
      
      Provide time constants in the module as they are useful to have and
      mirror constants provided by Go's time.
      
      Note: timers implementation is very suboptimal for now.
      9c260fde
    • Kirill Smelkov's avatar
      golang: Don't keep test programs inside golang/ dir · 6e3b3ff4
      Kirill Smelkov authored
      Python adds dirname of run program to sys.path . This way when
      golang_test_goleaked.py runs it can import modules located under golang/
      just by their name. Until now this was not noticed, but in the next
      patch we are going to add golang.time module and if test program is run
      with golang/ in sys.path just plain `import time` won't import time from
      stdlib and instead import time from golang/ .
      
      Such behaviour can be mitigated by doing `from __future__ import
      absolute_import` and we do that, including in golang_test_goleaked.py
      (see 81dfefa0 "*: __future__ += absolute_imports; Use unified __future__
      everywhere"). However that does not prevent modules - even modules from
      stdlib - who are doing `import time` and not doing future
      absolute_import to import golang's time instead of stdlib. For example
      on PyPy2 threading imports time and then the test breaks:
      
        Traceback (most recent call last):
          File "/home/kirr/src/tools/go/pygolang/.tox/pypy-thread/site-packages/golang/golang_test_goleaked.py", line 24, in <module>
            from golang import go, chan
          File "/home/kirr/src/tools/go/pygolang/.tox/pypy-thread/site-packages/golang/__init__.py", line 38, in <module>
            import inspect, threading, collections, random, sys
          File "/usr/lib/pypy/lib-python/2.7/threading.py", line 15, in <module>
            from time import time as _time, sleep as _sleep
          File "/home/kirr/src/tools/go/pygolang/.tox/pypy-thread/site-packages/golang/time.py", line 30, in <module>
            from golang import go, chan, select, default, nilchan, panic
        ImportError: cannot import name 'go'
      
      -> Move the test program into a directory different from golang/ to
      avoid this trap.
      6e3b3ff4
  8. 14 May, 2019 1 commit
    • Kirill Smelkov's avatar
      *: __future__ += absolute_imports; Use unified __future__ everywhere · 81dfefa0
      Kirill Smelkov authored
      - we are going to introduce golang.time, and from inside there without
        `from __future__ import absolute_imports` it won't be possible to import
        needed stdlib's time.
      
      - we were already doing `from __future__ import print_function`, but
        only in some files.
      
      -> It makes sense to apply updated __future__ usage uniformly.
      81dfefa0
  9. 10 May, 2019 1 commit
    • Kirill Smelkov's avatar
      fixup! pygolang v0.0.1 · 1172e39f
      Kirill Smelkov authored
      - correct CHANGELOG title levels: since readme uses "----" for the
        first level and "~~~~" for the second level, "====" was interpreted as
        the third title level and "Pygolang change history" became
        sub-sub-section of "Additional packages and utilities" with section
        for each version conversely becoming first level. It was not very
        noticeable until 0c5f9d06 (readme: Push "Additional packages and
        utilities" into its own section) started to use "~~~~".
        -> Rework CHANGELOG titling to align with the one in README.
      
      - fix minor markup bits in README.
      1172e39f
  10. 09 May, 2019 1 commit
  11. 03 May, 2019 3 commits
  12. 02 May, 2019 2 commits
    • Kirill Smelkov's avatar
      readme: Push "Additional packages and utilities" into its own section · 0c5f9d06
      Kirill Smelkov authored
      We are going to add more packages to mirror Go stdlib (e.g. next is
      context). Organize a dedicated section for what is additionally provided
      besides mirroring Go language and refer to that place only once from the
      top-level abstract.
      0c5f9d06
    • Kirill Smelkov's avatar
      golang: Add support for nil channel · 2aad64bb
      Kirill Smelkov authored
      Send/recv on the nil channel block forever; close panics.
      If a nil channel is used in select - corresponding case is never selected.
      
      Setting channel to nil is a usual idiom in Go to disable processing some
      cases in select. Nil channel is also used as "done" for e.g.
      context.Background() - for contexts that can be never canceled.
      2aad64bb
  13. 24 Mar, 2019 3 commits
    • Kirill Smelkov's avatar
      golang: Kill @method · 262f8986
      Kirill Smelkov authored
      It was deprecated and scheduled to be removed in 942ee900 (golang:
      Deprecate @method(cls) in favour of @func(cls)).
      262f8986
    • Kirill Smelkov's avatar
      pygolang v0.0.0.dev8 · 0f60aa77
      Kirill Smelkov authored
      0f60aa77
    • Kirill Smelkov's avatar
      gpython: Workaround PyPy2 vs Gevent patch_thread crash · 704d99f0
      Kirill Smelkov authored
      	pypy-gevent runtests: commands[0] | gpython -m pytest gpython/ golang/
      	Traceback (most recent call last):
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/bin/gpython", line 10, in <module>
      	    sys.exit(main())
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/site-packages/gpython/__init__.py", line 153, in main
      	    _ = monkey.patch_all()      # XXX sys=True ?
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/site-packages/gevent/monkey.py", line 976, in patch_all
      	    patch_thread(Event=Event, _warnings=_warnings)
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/site-packages/gevent/monkey.py", line 178, in ignores
      	    return func(*args, **kwargs)
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/site-packages/gevent/monkey.py", line 588, in patch_thread
      	    _patch_existing_locks(threading_mod)
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy-gevent/site-packages/gevent/monkey.py", line 492, in _patch_existing_locks
      	    if o._RLock__owner is not None:
      	AttributeError: 'thread.RLock' object has no attribute '_RLock__owner'
      
      It will be fixed on next Gevent release, however until then the crash is there
      and we have to workaround: we can skip patching existing locks as it will be
      the same behaviour of next Gevent release, since its just not possible to patch
      locks created via instantiated C-level classes:
      
      https://github.com/gevent/gevent/commit/d0e04658
      
      We are doing monkey-patching very early, so it should be safe.
      704d99f0
  14. 20 Mar, 2019 1 commit
    • Kirill Smelkov's avatar
      golang: Deflake channel tests · 731f39e3
      Kirill Smelkov authored
      Channel tests are passing on an idle machine. However if there is
      another load - e.g. in the presence of simultaneous Firefox start,
      channel tests break, for example:
      
      	    def test_select():
      	        ...
      
      	        # non-blocking try recv: ok
      	        ch = chan()
      	        done = chan()
      	        def _():
      	            for i in range(N):
      	                ch.send(i)
      	            done.close()
      	        go(_)
      
      	        for i in range(N):
      	            tdelay()
      	            if i % 2:
      	                _, _rx = select(
      	                        ch.recv,
      	                        default,
      	                )
      	                assert (_, _rx) == (0, i)
      	            else:
      	                _, _rx = select(
      	                        ch.recv_,
      	                        default,
      	                )
      	>               assert (_, _rx) == (0, (i, True))
      	E               assert (1, None) == (0, (320, True))
      	E                 At index 0 diff: 1 != 0
      	E                 Use -v to get the full diff
      
      	golang_test.py:209: AssertionError
      
      The failure here is that it was default case selected, not ch.recv_. The
      default was selected because the sending goroutine was not fast to
      enqueue next send before we tried to receive. We were trying to make
      sure that the sender will be enqueued first via adding 1ms time delay
      before trying to receive, but in the presence of concurrent load spikes
      that turns out to be not enough.
      
      We could try to fix the test by increasing the time to sleep in tdelay,
      make the tests more slow and still not 100% reliable. However we can
      change the tests to instead actually wait for the condition that is
      semantically required: a sender enqueued on the channel.
      
      Do that everywhere where tdelay was used.
      
      Now tests are faster (it was ~3s total, now it is ~ 0.5s total) and pass
      irregardless of whether the machine is idle or otherwise loaded.
      731f39e3
  15. 19 Mar, 2019 3 commits
    • Kirill Smelkov's avatar
      tox -= Python3.6 · 469f21a9
      Kirill Smelkov authored
      Debian dropped Python3.6 from python3-distutils. Without it I'm no
      longer being able to run tests on that python version:
      
      	(neo) (z-dev) (g.env) kirr@deco:~/src/tools/go/pygolang$ tox -e py36-thread
      	execution failed:  -- Traceback (most recent call last):
      	  File "<string>", line 1, in <module>
      	ModuleNotFoundError: No module named 'distutils.sysconfig'
      
      	GLOB sdist-make: /home/kirr/src/tools/go/pygolang/setup.py
      	py36-thread create: /home/kirr/src/tools/go/pygolang/.tox/py36-thread
      	ERROR: invocation failed (exit code 1), logfile: /home/kirr/src/tools/go/pygolang/.tox/py36-thread/log/py36-thread-0.log
      	ERROR: actionid: py36-thread
      	msg: getenv
      	cmdargs: '/usr/bin/python3 -m virtualenv --python /usr/bin/python3.6 py36-thread'
      
      	Traceback (most recent call last):
      	  File "/usr/lib/python3/dist-packages/virtualenv.py", line 25, in <module>
      	    import distutils.sysconfig
      	ModuleNotFoundError: No module named 'distutils.sysconfig'
      	Running virtualenv with interpreter /usr/bin/python3.6
      
      	ERROR: Error creating virtualenv. Note that some special characters (e.g. ':' and unicode symbols) in paths are not supported by virtualenv. Error details: InvocationError('/usr/bin/python3 -m virtualenv --python /usr/bin/python3.6 py36-thread (see /home/kirr/src/tools/go/pygolang/.tox/py36-thread/log/py36-thread-0.log)', 1)
      
      https://bugs.debian.org/918881#19
      469f21a9
    • Kirill Smelkov's avatar
      gpython: Don't check for time not being pre-imported on PyPy · e847c550
      Kirill Smelkov authored
      PyPy7 always pre-imports the time module. Without the change tests were
      failing:
      
      	Traceback (most recent call last):
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy3-gevent/bin/gpython", line 10, in <module>
      	    sys.exit(main())
      	  File "/home/kirr/src/tools/go/pygolang/.tox/pypy3-gevent/site-packages/gpython/__init__.py", line 145, in main
      	    '\n\n\t%s\n\nsys.modules:\n\n\t%s' % (bad, sysmodv))
      	RuntimeError: gpython: internal error: the following modules are pre-imported, but must be not:
      
      	        ['time']
      
      	sys.modules:
      
      	        ['__future__', '__main__', '__pypy__', '__pypy__._pypydatetime', '__pypy__.builders', '__pypy__.intop', '__pypy__.os', '__pypy__.thread', '__pypy__.time', '_ast', '_bootlocale', '_codecs', '_collections', '_collections_abc', '_continuation', '_csv', '_frozen_importlib', '_frozen_importlib_external', '_imp', '_io', '_locale', '_multibytecodec', '_operator', '_rawffi', '_rawffi.alt', '_signal', '_sre', '_structseq', '_thread', '_warnings', '_weakref', '_weakrefset', 'abc', 'array', 'builtins', 'codecs', 'copyreg', 'encodings', 'encodings.aliases', 'encodings.ascii', 'encodings.latin_1', 'encodings.utf_8', 'errno', 'gc', 'genericpath', 'gpython', 'marshal', 'os', 'os.path', 'posix', 'posixpath', 'pwd', 're', 'site', 'sre_compile', 'sre_constants', 'sre_parse', 'stat', 'sys', 'time', 'unicodedata']
      e847c550
    • Kirill Smelkov's avatar
      tox += PyPy3 · da68a8ae
      Kirill Smelkov authored
      Debian started to ship PyPy3:
      
      	https://packages.debian.org/search?keywords=pypy3
      da68a8ae