View source
py3.5-scandir
  • Contributors
    1. Loading...
Author Commit Message Date Builds
250+ commits behind default.
Armin Rigo
DirEntry.__repr__()
Armin Rigo
inode()
Armin Rigo
hg merge py3.5
Armin Rigo
re.compile().__repr__
Armin Rigo
update
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
re.fullmatch()
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
Improve testing of validate.py. Fix it.
Armin Rigo
fix test
Armin Rigo
Finally found and fixed the UnicodeDecodeError escaping in RPython
Armin Rigo
Fix test
Armin Rigo
Fix for test_kw_defaults_None
Armin Rigo
translation fix
Armin Rigo
Be more careful about encoding/decoding to utf-8, as even the .encode('utf-8') crashes if given the unichar that is normally used for surrogates
Armin Rigo
Test and fix (and remove the temporary dirty fix which only works if there is a list object around)
Armin Rigo
Hack at importlib/_bootstrap_external.py to use the same magic version number for pyc files as the one declared in pypy/interpreter/pycode.py.
Philip Jenvey
merge default
Philip Jenvey
fix: save the handles for CloseHandle
Ronan Lamy
hg merge py3k
Ronan Lamy
hg merge default
Armin Rigo
hg merge py3.5-marshal3
Armin Rigo
close branch, ready to merge
Armin Rigo
hg merge py3.5
Armin Rigo
translation fixes
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default (in particular, 9332dfa6e22e does not apply any more)
Philip Jenvey
work on the handle
Philip Jenvey
fix translation (win32)
Philip Jenvey
missing include (osx)
Ronan Lamy
Resynchronise rpython/ with branch 'py3.5'
Armin Rigo
hg merge py3.5
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
Write a version-dependent cache here
Armin Rigo
Bump the magic number, to avoid using old .pyc files
Armin Rigo
fix
Armin Rigo
Fixing marshalling of code objects as an attempt to be compatible with CPython.
Armin Rigo
more test fixes
Armin Rigo
hg merge py3.5
Armin Rigo
test fix
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
in-progress
Armin Rigo
in-progress
Armin Rigo
hg merger py3k
Armin Rigo
hg merge default
Armin Rigo
hg merge py3.5-noninherit Newly created file descriptors are non-inheritable (PEP 446)
Armin Rigo
close branch, ready to merge
Armin Rigo
fix test
Armin Rigo
Fixes in the test of fcntl
Armin Rigo
fix test
Armin Rigo
Marshal version 3
Armin Rigo
Write down the possible solution, to do later
Armin Rigo
List the functions here
Armin Rigo
Translation fixes
Armin Rigo
hg merge py3.5
Armin Rigo
small fix to hex()
Armin Rigo
hg merge py3k
Armin Rigo
memoryview.setitem()
Armin Rigo
Backport memoryview.cast(), fix getitem
Ronan Lamy
Restore support for the memoryview attributes that were already implemented
Ronan Lamy
Fix memoryview.__setitem__
Armin Rigo
some XXXs
Armin Rigo
copy the logic from 3.5 to inherit some file descriptors
Armin Rigo
_socket.dup(). and _socket.fromfd()... which no longer exists at interp-level, easy
Armin Rigo
hg merge py3.5
Armin Rigo
add trailing comma
Armin Rigo
hg merge py3.5
Armin Rigo
hg merge py3k
Armin Rigo
test and fix for _socket.dup()
Armin Rigo
socket.accept()
Armin Rigo
Cache the ENOSYS error from pipe2
Armin Rigo
add a test (passing)
Armin Rigo
socket.socket()
Armin Rigo
Add rsocket.socketpair(inheritable=..). Fix _socket.socketpair() at app-level.
Armin Rigo
select.devpoll()... is not implemented so far, but write it down anyway
Armin Rigo
Catch OSErrors
Armin Rigo
select.kqueue(), can't test but simple enough
Armin Rigo
select.epoll()
Armin Rigo
os.openpty()
Armin Rigo
os.fdopen()
Armin Rigo
os.open()
Armin Rigo
re-raise unlikely exception at app-level
Armin Rigo
io.open()
Armin Rigo
FileIO.__init__() needs to close the previous fd, if any
Armin Rigo
hg merge py3.5
Armin Rigo
Fix: the cache didn't work on py3.5, because we get two different calls
Armin Rigo
Fail if we call an llexternal() function with more arguments than declared
Armin Rigo
os.dup(), os.dup2()
Armin Rigo
Make pipe() return non-inheritable fds using the now-official way. Add pipe2().
Armin Rigo
Implement the basics
Armin Rigo
Remove file, probably added during a merge by mistake
Armin Rigo
pep446: Make newly created file descriptors non-inheritable
Philip Jenvey
cleanup
Ronan Lamy
Revert the main part of 26dfa2f67afd since it breaks too many random things, e.g. it caused pypy/module/imp/test/test_import.py::AppTestMultithreadedImp.test_multithreaded_import to hang.
Armin Rigo
Merged in marky1991/pypy_new/py3k_add_terminal_size (pull request #475) Add os.get_terminal_size (py3k)
Mark Young
Fix translation.
Mark Young
Whitespace fix.
Mark Young
Slight tweaks
Mark Young
get_terminal_size lib-python tests pass on windows after translation.
Mark Young
lib-python test passes for linux now at least.
Mark Young
Posting what I have so far.
Richard Plangger
merge py3k
Richard Plangger
revert last commit, remove already defined function in RPython (is provided in C)
Ronan Lamy
sys_module.filesystemencoding is not wrapped
Richard Plangger
remove double definition of cpyext C symbol
Richard Plangger
missing import
Richard Plangger
missing import
Richard Plangger
merge py3k
Richard Plangger
buffer protocal c headers are stripped down in py3
Richard Plangger
import issue
Armin Rigo
Merged in marky1991/pypy_new/py3k (pull request #474) Define clock_t on windows (py3k)
Mark Young
Merging with upstream 3k
Mark Young
Merging from windows with linux side
Mark Young
Define clock_t on windows too.
Armin Rigo
Update the comments about the reason for duplicating these few lines of code from lib-python/3/code.py.
Richard Plangger
cpyext, struct to usemodules
Richard Plangger
pull some merge conflicts from py3k
Richard Plangger
resolve some merge conflicts
Richard Plangger
merge py3k
Richard Plangger
merge default
Ronan Lamy
hg merge default
Raffael Tfirst
Describe key-value reverse order bug for dicts
Raffael Tfirst
Fix unpack_map bug saying X is not a mapping if another element is unpacked before (reset is_unpacking to False if element doesn't have to be unpacked)
Raffael Tfirst
Change error message on unpack and extended unpack if there are more values expected than available
Armin Rigo
Reduce the diff, don't import cpyext in this _cffi_backend test
Armin Rigo
hg merge default
Armin Rigo
hg merge default
Armin Rigo
Win32: don't crash translation
Armin Rigo
translation fix
Armin Rigo
Fix for posix/test/test_ztranslation
Armin Rigo
fix test
Armin Rigo
hg merge py3.5-scandir
Armin Rigo
ready for merge
Armin Rigo
Fix os.scandir(None), and os.scandir(dirfd) does not work on cpython either
Armin Rigo
hg merge py3.5
Armin Rigo
hg merge py3k
Armin Rigo
os.listdir(None)
Armin Rigo
Test and fix for os.listdir(fd)
Armin Rigo
hg merge default
Armin Rigo
Protect against usage in multiple threads
Armin Rigo
Finish the first draft of os.scandir() (missing: Win32)
Armin Rigo
hg merge default
Armin Rigo
in-progress: is_file() etc.
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
Continue working on os.scandir() on a branch
Armin Rigo
No point in making 'name' lazy, and it's better if the EncodingError we might get is thrown at the same place as in CPython
Armin Rigo
Returning a choice of bytes or unicodes
Armin Rigo
Start on posix.scandir()
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
hg merge py3k
Armin Rigo
translation fix
Armin Rigo
Eh, I don't know how I missed that __kwdefaults__ could simply be a moduledict
Armin Rigo
hg merge py3k
Armin Rigo
hg merge py3k-kwonly-builtin Fix kwonly handling for built-in functions, and also probably fix the handling of kwonly arguments in general in the JIT
Armin Rigo
close branch, ready to merge
Armin Rigo
Put the ".decode('utf-8')" here, whenever we wrap an RPython-level string as an app-level unicode attribute name
Armin Rigo
hg merge py3k
Armin Rigo
Add an assert that we get consistently the rpython 'str' type here, and fix two tests that passed 'unicode' explicitly
Armin Rigo
A failing test, which fails for the same reason as test_function.test_kwdefaults does
Armin Rigo
hg merge py3k
Armin Rigo
oops, broke translation
Armin Rigo
Test and fix
Armin Rigo
fix test, remove unused method
Armin Rigo
Tests, fix
Armin Rigo
fixes
Armin Rigo
in-progress
Armin Rigo
Fix for test_close_on_collect
Armin Rigo
hg merge py3k
Armin Rigo
Remove the "from __future__ import xyz" from the app-level helpers: it has no effect any more, and it loads the world by using importlib
Armin Rigo
A much simpler solution gets us most of the way there (thanks fijal): rely on the mapdict logic
Armin Rigo
Backed out changeset b130c20feba7 Not so easy: it wouldn't support ``f.__kwdefaults__['a']=5`` or logic like ``f.__kwdefaults__=d={}; d['a']=5``.
Armin Rigo
more fixes
Armin Rigo
merge heads
Armin Rigo
in-progress
Armin Rigo
A branch to implement kwonly specification for builtin functions
Armin Rigo
Another important speed-up for space initialization: instead of compiling importlib/_bootstrap.py every time, marshal the bytecode and cache it in rpython/_cache/.
Armin Rigo
Load sys.filesystemencoding and the app-level sys.std{in,out,err} lazily. This gives a big reduction over the time it takes to build a space.
Armin Rigo
I *think* this expresses the logic that we want more cleanly
Raffael Tfirst
Change const.value to const.obj in test_obj because Const in AST changed attribute name
Ronan Lamy
Backout 7b97d4152ebc: pypy behaves like CPython wrt. .pyo now
Ronan Lamy
Allow lone surrogates in _pypyjson
Ronan Lamy
hg merge py3k
Ronan Lamy
hg merge default
Armin Rigo
hg merge py3k
Armin Rigo
hg merge default
Armin Rigo
Describe more clearly that the (first) StackDepthComputationError can be ignored at first.
Armin Rigo
Very obscure fix
Raffael Tfirst
Change POP_EXCEPT back to -1
Raffael Tfirst
Merge with py3.5-async
Raffael Tfirst
Check for correct values in 'async for' test
Armin Rigo
Manual copy of b4e83acb5f3c, and fix of a resulting issue: two opcodes had a wrong stack effect
Ronan Lamy
Fix constant-folding for '**'
Ronan Lamy
hg merge py3k
Armin Rigo
Seems that sys.__package__ is also '' on CPython 3.5.
Armin Rigo
Add the new RecursionError exception.
Ronan Lamy
hg merge default
Armin Rigo
frame.clear()
Raffael Tfirst
Change stack effect of POP_EXCEPT from -1 to -2, to compensate the depth in 'async for', because all except-finally depths are different in cpython. shorten the 'async for' test case by 5 seconds
Raffael Tfirst
Change POP_TOP to DUP_TOP in async_for, should fix TypeError after execution of 'async for'
Richard Plangger
missing argument to operation error
Richard Plangger
32bit translation issue, prebuilt long
Richard Plangger
hex method for bytes and bytearray + test
Richard Plangger
hex method for memoryview + tests
Raffael Tfirst
Minor change in TODO for __anext__
Raffael Tfirst
Add TODO for __anext__ in coroutine wrapper
Raffael Tfirst
Change __next__ to __anext__ in coroutine
Raffael Tfirst
Add lookup for __anext__ in next() and __anext__ to typedef of coroutines, add better (offline) async_for test, remove old async_for test and change it back to test only the gil error
Armin Rigo
Fix the test list, and skip for now some that seem to deadlock
Raffael Tfirst
Add StopAsyncIteration exception, expand test to check 'async for' in test_asyncio
Armin Rigo
missing import
Armin Rigo
Allow (and ignore) duplicate names in a single __slots__ specification. Believe it or not, the stdlib datetime.py does that.
Armin Rigo
Quick fix, hopefully fixing issues of subprocess.Popen() in a translated pypy
Armin Rigo
Add rposix.{get,set}_inheritable(), needed by Python 3.5
Armin Rigo
slice.indices() is rewritten, duplicating functionality instead of reusing it
Armin Rigo
fix docstring
Armin Rigo
_imp.load_dynamic => _imp.create_dynamic
Raffael Tfirst
Add description to test_asynchronous_context_managers
Raffael Tfirst
Fix async with test, coroutines can assign lock in any order, but release of one lock has to appear before hold of the other lock
Raffael Tfirst
Fix async with test, fixes timing for messages expected in assert
Raffael Tfirst
Merge with py3.5
Armin Rigo
Revert all changes to this from Python 3.5.1. This mainly reverts 1341a432e134, which I think is not needed any more in 3.5
Armin Rigo
fixes
Armin Rigo
Fix site.py
Ronan Lamy
Add usemodules declarations that py3k seems to require
Raffael Tfirst
Add asyncio test for asynchronous context managers
Ronan Lamy
hg merge default
Raffael Tfirst
Merge with py3k
Raffael Tfirst
Undo change of conftest (load python3 again), needs more fixes later
Raffael Tfirst
Change python environment in conftest to python3.5
Ronan Lamy
Fix bad change in b053ff5c2d6d
Ronan Lamy
Fix translation
Ronan Lamy
Remove code added by py2-specific branch 'resource_warning'
Raffael Tfirst
Fix BUILD_SET_UNPACK by changing iterator to iter instead of itervalues as w_item should never be a dict anyway
Manuel Jacob
hg merge py3k
Raffael Tfirst
Fix validate tests, ast classes "arguments", "classdef" and "call" don't have stararg and kwarg anymore
Raffael Tfirst
Merge with py3.5-async
Raffael Tfirst
Fix compiler test (positional arguments can follow starred arguments)
Raffael Tfirst
Fix and rename test_crap_after_starargs (allowed in PEP 448), dirty fix in function calls if argument order is reversed (happens if stararg occurs after kwarg)
Ronan Lamy
Merged in marky1991/pypy_new/py3k (pull request #468) Py3k: Fix Translation for FreeBSD
Mark Young
Reapply the fix from issue 2348. Translation works now.
Mark Young
Reduce diff with cpython. (thanks for pointing this out, David\!)
Mark Young
Fixed translation for freebsd.
Mark Young
Mostly fix translation on freebsd. A failure still happens late in translation, but rtyping works at least.
Mark Young
Merging.
4440 commits not shown.