[pypy-commit] [Git][pypy/pypy][branch/py3.6] 42 commits: (antocuni, arigo): add a passing test which we needed to convince ourselves that…

Matti Picus foss at heptapod.net
Wed Mar 18 07:40:29 EDT 2020


Matti Picus pushed to branch branch/py3.6 at PyPy / pypy


Commits:
9d641148 by Antonio Cuni at 2019-11-18T15:27:18+01:00
(antocuni, arigo): add a passing test which we needed to convince ourselves that you can raise exceptions from within llhelper()ed functions

--HG--
branch : hpy-rpython-backports

- - - - -
a5621b11 by Armin Rigo at 2019-11-18T18:38:29+01:00
(antocuni, arigo)

Allow llhelper functions to propagate exceptions from RPython to RPython via C.
This already works after translation, but we need to convince ll2ctypes.

--HG--
branch : hpy-rpython-backports

- - - - -
51d004db by Antonio Cuni at 2019-11-18T23:24:43+01:00
introduce a new decorator @llhelper_can_raise and use it to fix test_exception

--HG--
branch : hpy-rpython-backports

- - - - -
fb42d8b2 by Armin Rigo at 2020-03-07T18:53:39+01:00
restore _immutable_fields_ on the ExecutionContext class, which was removed in b37f9be92e79 but is probably still useful

- - - - -
e80f30af by Manuel Jacob at 2020-03-08T03:22:27+01:00
Make build_cffi_imports.py work in environments that don’t have _multiprocessing.

- - - - -
48fd242b by Carl Friedrich Bolz-Tereick at 2020-03-08T12:18:16+01:00
a branch to try to help the jit reason about nested scopes better, Cells in
particular

--HG--
branch : nested-scopes-jit

- - - - -
d509474d by Manuel Jacob at 2020-03-08T12:57:31+01:00
Let pytest.py run with python2 by default.

- - - - -
a5f561a3 by Armin Rigo at 2020-03-08T13:10:25+01:00
merge heads

- - - - -
cb4e693f by Armin Rigo at 2020-03-08T12:41:31+00:00
Merge branch 'topic/default/make_build_cffi_imports_work_without__multiprocessing' into 'branch/default'

Make build_cffi_imports.py work in environments that don’t have _multiprocessing.

See merge request pypy/pypy!712

- - - - -
201d554e by Matti Picus at 2020-03-08T16:08:48+02:00
Update release note and fix some warnings

- - - - -
f3083f09 by Manuel Jacob at 2020-03-08T15:36:08+01:00
Add test for getitem slowpath.

- - - - -
6605958d by Manuel Jacob at 2020-03-08T16:08:26+01:00
Avoid using module name 'buffer' because it breaks app tests.

- - - - -
2c81935e by Manuel Jacob at 2020-03-08T16:15:38+01:00
Skip test that doesn’t make sense under runappdirect.

- - - - -
0f6ac5a7 by Manuel Jacob at 2020-03-08T19:19:11+01:00
Make write() method work with memoryview.

On my machine under Python 2.7.17, the test failed previously because
io.BufferedWriter passes a memoryview into the raw stream’s write() method.

I verified that the previous test failure is unrelated to my recent changes.

- - - - -
17d62954 by Manuel Jacob at 2020-03-08T19:22:06+01:00
Make check of exception string more generic for different Python implementations.

- - - - -
fe216fc2 by Carl Friedrich Bolz-Tereick at 2020-03-09T14:35:53+01:00
start implementing Cell families, that Cells are grouped into, according to
their outer defining function. Goal is to be able to track immutability of
Cells (like mapdict instance fields).

Doesn't work so far, since the initializing set is counted as a mutation so
far.

--HG--
branch : nested-scopes-jit

- - - - -
e1403f7f by Carl Friedrich Bolz-Tereick at 2020-03-09T17:35:16+01:00
fix the failing test: do the cell initialization differently, first initialize
only those cells that aren't arguments. After argument parsing, create the
cells with the correct values directly. That way the tracking whether cells are
ever mutated works correctly for arguments.

--HG--
branch : nested-scopes-jit

- - - - -
a137d056 by Carl Friedrich Bolz-Tereick at 2020-03-09T17:47:27+01:00
use one family per cell

--HG--
branch : nested-scopes-jit

- - - - -
1efa3acb by Carl Friedrich Bolz-Tereick at 2020-03-09T18:05:48+01:00
fix bug

(also, it helps to add the test file)

--HG--
branch : nested-scopes-jit

- - - - -
02289b38 by Manuel Jacob at 2020-03-10T19:22:11+01:00
Add xfailing test for __pypy__.bytebuffer.__getslice__() with start > stop.

- - - - -
a2876376 by Carl Friedrich Bolz-Tereick at 2020-03-11T20:29:07+01:00
fix test_generators

the two extra setfields that look scary aren't actually new in the trace, they
are just moved around. They are forced by the following guard_not_invalidated,
which moved around in the trace

- - - - -
8dd939c5 by Antonio Cuni at 2020-03-13T11:47:37+01:00
close branch

--HG--
branch : hpy-rpython-backports

- - - - -
0611ef3d by Antonio Cuni at 2020-03-13T11:48:54+01:00
Transplant to default some rpython/ commits which were done in the hpy
branch. The original hpy commits are:

3a384fd65d2e
1b05295469eb
87553600e78a

- - - - -
586265e8 by Armin Rigo at 2020-03-13T12:15:05+01:00
Change the signature of rlib.buffer.Buffer.getslice():

Now it doesn't take a 'stop' argument any more.  This prevents
bugs in some subclasses' implmentations that rely on 'stop' to
be equal to 'start + step*size'.  This was not always true,
notably if 'size == 0'.

- - - - -
92277dd5 by Armin Rigo at 2020-03-13T13:01:10+01:00
merge heads

- - - - -
6f180f8d by Armin Rigo at 2020-03-13T13:07:27+01:00
Untranslated, rlib.rgil now implements its own "gil" instead of relying on the
one from the C code---which doesn't work because the latter falls back to
"1234" as the thread ident, even if there are multiple threads in the test

- - - - -
bd512511 by Manuel Jacob at 2020-03-13T13:56:46+01:00
Support buffer slices with the end index larger than the length of the buffer in RPython (in __getslice__()).

All other RPython classes seem to support that as well (e.g. rstr and rlist).

- - - - -
e5271a3b by Manuel Jacob at 2020-03-14T22:26:06+01:00
Make script work with new versions of Mercurial.

- - - - -
b27b0ea0 by Carl Friedrich Bolz-Tereick at 2020-03-16T13:45:08+01:00
merge default

--HG--
branch : nested-scopes-jit

- - - - -
3b16bb59 by Carl Friedrich Bolz-Tereick at 2020-03-16T13:48:31+01:00
fix test, this test shows the effect of the branch: global immutable cells can
have their access completely folded away

--HG--
branch : nested-scopes-jit

- - - - -
0a92bf79 by Carl Friedrich Bolz-Tereick at 2020-03-16T18:11:07+01:00
make the cells optimization also work for non-arguments, as long as the
variable is only written to once

--HG--
branch : nested-scopes-jit

- - - - -
2576be14 by Carl Friedrich Bolz-Tereick at 2020-03-16T23:27:00+01:00
undo most of d2497f614daa: simplify Cell initialization back to what it was on
default, now that Cell change tracking is somewhat more robust

--HG--
branch : nested-scopes-jit

- - - - -
2eb4875a by Carl Friedrich Bolz-Tereick at 2020-03-17T14:27:40+01:00
strengthen test, add a proper test for non-interference

--HG--
branch : nested-scopes-jit

- - - - -
ed9633e2 by Antonio Cuni at 2020-03-17T14:41:05+01:00
fix test_whatsnew (no need to document the branch, nothing relevant to whatsnew)

- - - - -
8ee78dde by Carl Friedrich Bolz-Tereick at 2020-03-17T15:18:36+01:00
document branch

--HG--
branch : nested-scopes-jit

- - - - -
8972f09b by Carl Friedrich Bolz-Tereick at 2020-03-17T15:19:37+01:00
close to-be-merged branch

--HG--
branch : nested-scopes-jit

- - - - -
e04f626f by Carl Friedrich Bolz-Tereick at 2020-03-17T15:23:54+01:00
merge nested-scopes-jit

teach the JIT to reason better about nested scopes. In particular, track
whether the cells of a single local variable are ever mutated, and if they
aren't, constant fold the read access from them, if they are constant.

- - - - -
75d4e0a1 by Carl Friedrich Bolz-Tereick at 2020-03-17T15:25:31+01:00
merge heads

- - - - -
a10ccf6a by Carl Friedrich Bolz-Tereick at 2020-03-18T09:16:06+01:00
add two global functions that give information on socket fds that 3.7 needs

- - - - -
eaa366a7 by Matti Picus at 2020-03-18T11:43:08+02:00
maybe fix sysconfig.get_config_var('SOABI') ?

--HG--
branch : py3.6

- - - - -
9ef1dc52 by Matti Picus at 2020-03-18T11:56:25+02:00
merge default into branch

--HG--
branch : py3.6

- - - - -
5bdde103 by Matti Picus at 2020-03-18T13:36:44+02:00
fix merge

--HG--
branch : py3.6

- - - - -


30 changed files:

- lib_pypy/_sysconfigdata.py
- lib_pypy/tools/build_cffi_imports.py
- pypy/doc/__pypy__-module.rst
- pypy/doc/coding-guide.rst
- pypy/doc/man/pypy.1.rst
- pypy/doc/mercurial_heptapod.rst
- pypy/doc/release-v7.3.0.rst
- pypy/doc/release-v7.3.1.rst
- pypy/doc/tool/makecontributor.py
- pypy/doc/whatsnew-head.rst
- pypy/interpreter/nestedscope.py
- pypy/interpreter/pycode.py
- pypy/interpreter/pyframe.py
- + pypy/interpreter/test/test_cellfamily.py
- pypy/module/__pypy__/test/test_bytebuffer.py
- pypy/module/_cffi_backend/cbuffer.py
- pypy/module/_cffi_backend/func.py
- pypy/module/_io/test/test_bufferedio.py
- pypy/module/_multiprocessing/test/test_interp_semaphore.py
- pypy/module/_sre/interp_sre.py
- + pypy/module/_sre/test/__init__.py
- + pypy/module/_sre/test/test_sre_buf.py
- pypy/module/array/interp_array.py
- pypy/module/cpyext/buffer.py
- pypy/module/cpyext/test/test_pyerrors.py
- pypy/module/marshal/interp_marshal.py
- pypy/module/mmap/interp_mmap.py
- pypy/module/pypyjit/test_pypy_c/test_call.py
- pypy/module/pypyjit/test_pypy_c/test_generators.py
- pypy/module/struct/formatiterator.py


View it on GitLab: https://foss.heptapod.net/pypy/pypy/compare/2cf0d1f15e64e02d7ad431582fb4ab2273472697...5bdde103aa223e6a0ea6c84680bb76f873ae3531

---
View it on GitLab: https://foss.heptapod.net/pypy/pypy/compare/2cf0d1f15e64e02d7ad431582fb4ab2273472697...5bdde103aa223e6a0ea6c84680bb76f873ae3531
You're receiving this email because of your account on foss.heptapod.net.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/pypy-commit/attachments/20200318/a361206b/attachment-0001.html>


More information about the pypy-commit mailing list