Branches
wrap-bytes
-
- Loading...
| Author | Commit | Message | Date | Builds | ||
|---|---|---|---|---|---|---|
| 250+ commits behind default. | ||||||
|
a160002
M
|
hg merge SomeString-charclass
|
|
|||
|
|
Improve test, and better "improve" the merge of AsciiString with SomeChar.
|
|
||||
|
|
a5ac58f
M
|
hg merge SomeString-charclass
|
|
|||
|
|
Fix, and add support for "x in 'ascii_str'".
|
|
||||
|
|
Oops2
|
|
||||
|
|
oops
|
|
||||
|
|
65a411f
M
|
hg merge SomeString-charclass, directly into py3k
|
|
|||
|
|
Add Utf8Char kind.
Will be used by space.identifier_w()
|
|
||||
|
|
Progress, add AsciiChar kind.
|
|
||||
|
|
Replace no_nul by the more general "character kind".
|
|
||||
|
|
A branch to experiment with "character class", annotation about the content of a string
|
|
||||
|
|
A branch to experiment with space.wrap(bytes): only accept proven ascii strings
|
|
||||
|
|
issue1797: fix get_python_lib(standard_lib=True) per our 'lib-python/3' layout
|
|
||||
|
|
Backed out changeset 62fc05632b25: wrong branch
|
|
||||
|
|
switch to unicodedata 6.2.0
|
|
||||
|
|
unneeded import
|
|
||||
|
|
avoid triggering the now lazy __context__ setup 'up front' (when grabbing it
from the current frame) when breaking __context__ chain cycles, for now. this
is a trade-off: we won't break some cycles in obscure situations for the sake
of not paying a cost in probably more situations
|
|
||||
|
|
improve __context__ setup vis-a-vis the JIT:
defer its setup, when possible, until __context__ is explicitly requested by
searching for it through the traceback. it's not possible to defer when
__context__ comes from the current frame as the new exception eventually
overwrites it (in frame.last_exception)
in that case we do the setup sooner, when recording the traceback, with little
cost.. until we need to break __context__ chain cycles (wh…
|
|
||||
|
|
specialize range iterators when the ranges fit into machine sized integers and
further specialize when step is not specified (like the default branch does)
|
|
||||
|
|
w_kw_defs is quasi-immutable
|
|
||||
|
|
prefer finditem_str
|
|
||||
|
|
reduce diff w/ default
|
|
||||
|
|
_utf8 is quasi-immutable
|
|
||||
|
|
prefer find/setitem_str shortcuts
(grafted from f48daf83db6a568e814467d1e2132ee8d046b1d3)
|
|
||||
|
|
now an IOError subclass
|
|
||||
|
|
adapt to py3 and add get/setdefault
|
|
||||
|
|
gdbm -> _gdbm
|
|
||||
|
|
gdbm -> _gdbm
|
|
||||
|
|
ignore the py3k branch
|
|
||||
|
|
e72bf75
M
|
merge default
|
|
|||
|
|
adapt to py3
|
|
||||
|
|
fix the import
|
|
||||
|
|
have 'execfile' accept null bytes
|
|
||||
|
|
5a73bbc
M
|
merge default
|
|
|||
|
|
8f3c7c1
M
|
merge default
|
|
|||
|
|
re-enable the missed BytesListStrategy
|
|
||||
|
|
adapt to py3
|
|
||||
|
|
kill windows' popen funcs
|
|
||||
|
|
make our import tag prefix 'pypy3' to avoid clashing w/ pypy2 cffi's compiled
libs within __pycache__
|
|
||||
|
|
_winreg -> winreg
|
|
||||
|
|
fix windows' translation
|
|
||||
|
|
hopefully fix translation
|
|
||||
|
|
forgot to re-enable set's ByteStrategy, fix
|
|
||||
|
|
fix space.hash to properly handle 'longs'
|
|
||||
|
|
accommodate struct.error instead of TypeError now
|
|
||||
|
|
thread -> _thread
|
|
||||
|
|
skip these when hosted on a narrow build, they require a fully fledged wide build
|
|
||||
|
|
assume narrow build behavior here when hosted on a narrow python
|
|
||||
|
|
oops, avoid we_are_translated at module scope
|
|
||||
|
|
also handle surrogates when hosted on a narrow build
|
|
||||
|
|
skip when the fsencoding can't handle this filename
|
|
||||
|
|
ignore LocaleErrors
|
|
||||
|
|
2to3
|
|
||||
|
|
attempt to get more debug output
|
|
||||
|
|
Fix an obscure translation issue,
maybe because runicode functions are called with two kinds of strings.
|
|
||||
|
|
2032fc1
M
|
hg merge default
|
|
|||
|
|
f481a6c
M
|
hg merge default
|
|
|||
|
|
workaround a py3k appdirect failure
|
|
||||
|
|
disable for now, it's causing obscure crashes:
https://bitbucket.org/pypy/pypy/issue/1773
|
|
||||
|
|
fix potential UnicodeEncodeErrors under pytest.py --resultlog: the captured
output may come in as unicode
(grafted from ef6d8454c0a6cd10097c82056727f5cef8b7830e)
|
|
||||
|
|
update the sre support code to CPython 3.2.5's
|
|
||||
|
|
fix cpyext's version
|
|
||||
|
|
fix handling of EnvironmentError filenames
|
|
||||
|
|
fix jit translation, now that elidable is stricter
|
|
||||
|
|
handle new wide build behavior per 5494a374d576
|
|
||||
|
|
sync w/ default
|
|
||||
|
|
match cpython's error message
|
|
||||
|
|
switch to py3's error message
|
|
||||
|
|
f7d70d5
M
|
merge default
|
|
|||
|
|
adapt d083e472a6ab to py3k: treat get_cleared_operation_error as if it was no
exception here too
|
|
||||
|
|
py3 hasattr goes through getattr
|
|
||||
|
|
2to3
|
|
||||
|
|
fix ztranslations
|
|
||||
|
|
tweak the test runner's custom AssertionError to pass
test_exc.test_doc_and_module
|
|
||||
|
|
__builtin__ -> builtins
|
|
||||
|
|
copy_reg is copyreg on py3k
|
|
||||
|
|
9401f74
M
|
merge default
|
|
|||
|
|
kill more __module__s
|
|
||||
|
|
9c2e85c
M
|
merge default
|
|
|||
|
|
d8096ef
M
|
merge default
|
|
|||
|
|
disallow unicode
|
|
||||
|
|
fix memoryview.readonly degrading to int from bool
|
|
||||
|
|
skip this for now as it's also a problem on CPython, albeit not as easy to
reproduce on there
|
|
||||
|
|
use the new buffer interface, adapt tests
|
|
||||
|
|
adapt tests from default
|
|
||||
|
|
adjust expected per py3k
|
|
||||
|
|
kill long literal
|
|
||||
|
|
01e1b2d
M
|
merge upstream
|
|
|||
|
|
1abe697
M
|
merge default
|
|
|||
|
|
fix translation
|
|
||||
|
|
00ddc2f
M
|
merge default
|
|
|||
|
|
marshal.loads use buffers correctly
|
|
||||
|
|
fix test_bytesio
|
|
||||
|
|
kill bufferstr_or_u_w
|
|
||||
|
|
add getarg('y*'), use instead of bufferstr_w here
|
|
||||
|
|
redefine this
|
|
||||
|
|
kill off old buffer interface and its usage
|
|
||||
|
|
refix: disallow unicode
|
|
||||
|
|
adapt memoryview related tests to py3
|
|
||||
|
|
Buffers now require a readonly attribute
|
|
||||
|
|
|
|||||
|
|
2f47b24
M
|
merge default
|
|
|||
|
|
fix translation
|
|
||||
|
|
fix ztranslation
|
|
||||
|
|
seqiter is good enough for default so kill our AraryIterator
|
|
||||
|
|
c7e7aac
M
|
merge default
|
|
|||
|
|
port some improvements from default
|
|
||||
|
|
e43a6cc
M
|
merge upstream
|
|
|||
|
|
this is supported in py3k
|
|
||||
|
|
fix
|
|
||||
|
|
have array support the new buffer interface
|
|
||||
|
|
add space.w_memoryview
|
|
||||
|
|
fix fcntl
|
|
||||
|
|
buffer fixes for py3k unicode/bytes
|
|
||||
|
|
buffer_w now takes flags, Buffers must now be _immutable_
|
|
||||
|
|
|
|||||
|
|
fix imports
|
|
||||
|
|
rekill buffer and some py2 only tests
|
|
||||
|
|
67528c9
M
|
merge default (8a2e9e0c1676)
|
|
|||
|
|
092f39d
M
|
merge py3k-fix-strategies: re-enables the disabled strategies (issue1471)
|
|
|||
|
|
close to be merged branch
|
|
||||
|
|
utilize decodekey_str
|
|
||||
|
|
provide a listview_int for bytes
|
|
||||
|
|
provide the fromkeys fastpath, differently from default, since it's based
around unicode
|
|
||||
|
|
close it when finished
(grafted from 308ded060ac8ffbf1b1a16d88486558980b4bc1b)
|
|
||||
|
|
e0ce550
M
|
merge default (ea86924e88fb)
|
|
|||
|
|
revert back to usage of finditem/setitem_str
|
|
||||
|
|
remove extra double quote
|
|
||||
|
|
f3443cb
M
|
merge py3k
|
|
|||
|
|
23776e2
M
|
merge default
|
|
|||
|
|
73b4a5d
M
|
merge py3k
|
|
|||
|
|
act more like CPython's _PySequence_BytesToCharpArray: trigger MemoryErrors
with large __len__ results (unpackiterable doesn't). fixes
test_capi.test_seq_bytes_to_charp_array
|
|
||||
|
|
minor cleanup
|
|
||||
|
|
another bytes strat fix
|
|
||||
|
|
shortcut to bytes_w
|
|
||||
|
|
__builtin__ -> builtins
|
|
||||
|
|
a576218
M
|
merge default
|
|
|||
|
|
fix BytesStrategy iter to wrapbytes
|
|
||||
|
|
we now need bytes_w too
|
|
||||
|
|
more progress, have _str methods accept utf-8 encoded strs, for the sake of
compat. w/ default
|
|
||||
|
|
progress
|
|
||||
|
|
adapt disabled list strats
|
|
||||
|
|
fix cell/mapdict
|
|
||||
|
|
adapt tests, progress
|
|
||||
|
|
a0becb7
M
|
merge py3k
|
|
|||
|
|
f10a4d4
M
|
merge default
|
|
|||
|
|
a98806e
M
|
merge py3k
|
|
|||
|
|
a118df0
M
|
merge default
|
|
|||
|
|
2to3 the relative import
|
|
||||
|
|
8c349c4
M
|
merge default
|
|
|||
|
|
adapt http://hg.python.org/cpython/rev/f4271cc2dfb5 from cpython-3.3: we
already match 3.3's _sre.MAXREPEAT value for 32 bit, so we also need this
associated fix
|
|
||||
|
|
ebfddb6
M
|
merge py3k
|
|
|||
|
|
typo
|
|
||||
|
|
avoid running in an exception handler, since raising will chain exceptions
|
|
||||
|
|
help Popen find executable on win32
|
|
||||
|
|
update import library name on windows
|
|
||||
|
|
fix for import lib name (cc1896cf5625) and prepare import lib for tests (with TODO)
|
|
||||
|
|
do not rename windows import library in rpython, it belongs in pypy
|
|
||||
|
|
b143fe6
M
|
merge py3k
|
|
|||
|
|
fe69dee
M
|
merge default
|
|
|||
|
|
On Unicode wide builds (=all except win32), don't merge utf16 surrogate pairs on encoding.
This only affects python3 which sets allow_surrogates=False.
|
|
||||
|
|
copy import library pragma from cpython
|
|
||||
|
|
ec295db
M
|
merge default
|
|
|||
|
|
df78e44
M
|
merge py3k
|
|
|||
|
|
652e7bc
M
|
merge stdlib-3.2.5
|
|
|||
|
|
close to be merged branch
|
|
||||
|
|
another BufferError impl.
detail
|
|
||||
|
|
incoming paths are now unicode so their path manipulations should be too --
however rpath doesn't support unicode, so fsencode them for now
|
|
||||
|
|
911844b
M
|
merge py3k
|
|
|||
|
|
7eafa29
M
|
merge default
|
|
|||
|
|
712bbcb
M
|
merge py3k
|
|
|||
|
|
b706343
M
|
merge default
|
|
|||
|
|
use cpython 3.3's test which we already adhere to. 3.2.5's test seems bogus
(cpython issue15533)
|
|
||||
|
|
workaround for staticmethod descriptors on PyPy
|
|
||||
|
|
cpython issue12983: bytes literals with invalid \x escape now raise a
SyntaxError
|
|
||||
|
|
backout ebd9a9125c8c: len shouldn't raise a ValueError here. refs cpython
issue21173
|
|
||||
|
|
2f1ec8d
M
|
merge py3k
|
|
|||
|
|
dbedd0a
M
|
merge default
|
|
|||
|
|
Only run_fork_hooks('parent') if we called run_fork_hooks('before'),
otherwise we get strange errors about the Import lock
when a parameter does not have the expected type.
|
|
||||
|
|
Fix the test: fcntl is exactly the function I targeted when I changed space.c_filedescriptor_w().
|
|
||||
|
|
fix str() w/ out an object specified & kw only args
|
|
||||
|
|
CPython issue1692335: Exception.args is also set in __new__
(for naive exception subclasses which don't call super().__init__)
|
|
||||
|
|
we need a collect here
|
|
||||
|
|
reapply 4899736f152f to lib-python/3/ (no sys.getsizeof on pypy)
|
|
||||
|
|
Fix translation
|
|
||||
|
|
1984a39
M
|
hg merge py3k
|
|
|||
|
|
CPython 3.2.5 seems to be more careful when unwrapping file descriptors.
Do the same in pypy, and don't turn OverflowErrors into ValueErrors.
This will fix a failure in test_fcntl, but will probably break other tests. Need to watch buildbot.
|
|
||||
|
|
fix a c int overflow.
refs cpython issue15989
|
|
||||
|
|
Fix BytesSetStrategy: bytes items would be converted to unicode!
|
|
||||
|
|
b7947bb
M
|
merge py3k
|
|
|||
|
|
10edbab
M
|
merge default
|
|
|||
|
|
move a misplaced test
|
|
||||
|
|
exact error messages are an impl detail. we may revisit these in 3.4, when
cpython changes many of these messages, anyway
|
|
||||
|
|
Expat parser now correctly works with unicode input,
even when the XML internal encoding is not UTF8
(CPython issue 17089)
|
|
||||
|
|
mangle keyword-only argname when loading defaults (CPython issue #14607)
|
|
||||
|
|
Skip some implementation details.
One remaining failure in test_weakref is suspect: __len__() returns -7!
|
|
||||
|
|
pypy has no switchinterval and the test passes without.
|
|
||||
|
|
911b0a7
M
|
hg merge py3k
|
|
|||
|
|
branch for fixing/adapting the list/dict/etc strategies that were disabled
while focusing on py3k compat
|
|
||||
|
|
6c9441b
M
|
merge default
|
|
|||
|
|
Py3-ify some cpyext tests.
|
|
||||
|
|
170a321
M
|
merge default
|
|
|||
|
|
fix for 32bit platforms again
|
|
||||
|
|
adapt to py3, which allows bytes or unicode here
|
|
||||
|
|
e1951a6
M
|
merge default
|
|
|||
|
|
fix for 32bit platforms
|
|
||||
|
|
adapt to py3
|
|
||||
|
|
py3k still needs _check_released in buffer_w
|
|
||||
|
|
space.buffer was killed
|
|
||||
|
|
this now lives in baseobjspace after the buffer refactor
|
|
||||
|
|
restore our buffer related error message
|
|
||||
|
|
MiniBuffer's API actually differs from MemoryView, fix
|
|
||||
|
|
add a TODO comment
|
|
||||
|
|
adapt to buffer refactoring
|
|
||||
|
|
skip for now: requires the not NotImplemented audioop.lin2ulaw
|
|
||||
|
|
706e3a4
M
|
merge default
|
|
|||
|
|
6ff661c
M
|
merge default
|
|
|||
|
|
Fix translation.
|
|
||||
|
|
1cf13b1
M
|
merge default
|
|
|||
|
|
try to get away away with this simply subclassing MemoryView
|
|
||||
|
|
kill W_Buffer
|
|
||||
|
|
adapt to buffer refactor: not everything exposes a __buffer__ now
|
|
||||
|
|
kill __builtin__.buffer, adjust test
|
|
||||
|
|
fix translation
|
|
||||
|
|
fix/cleanup imports
|
|
||||
|
|
fb38194
M
|
merge default
|
|
|||
|
|
82be714
M
|
merge default
|
|
|||
|
|
add missing test entries
|
|
||||
|
|
update comment
|
|
||||
|
|
bump
|
|
||||
|
|
cbb08b3
M
|
merge vendor/stdlib
|
|
|||
|
|
25f71fd
M
|
merge default
|
|
|||
|
|
adjust the test as we follow 3.2.4? behavior now
|
|
||||
|
|
fix from last merge
|
|
||||
|
|
kill int.__div__
|
|
||||
|
|
workaround lack of space.hex/oct on py3k
|
|
||||
|
|
add int.__ceil/floor__, cleanup
|
|
||||
|
|
port skips from default
|
|
||||
|
|
50d4265
M
|
merge default
|
|
|||
|
|
8a77a2e
M
|
merge py3k-stdlib-2.7.6-merge
|
|
|||
|
|
close before merging
|
|
||||
|
|
2to3
|
|
||||
|
|
0913e40
M
|
merge default
|
|
|||
|
|
workaround this obscure py3k only failure
|
|
||||
|
|
adapt to py3
|
|
||||
|
|
woops
|
|
||||
|
|
adapt to py3
|
|
||||
|
|
2to3
|
|
||||
|
|
512fc23
M
|
merge default
|
|
|||
|
|
5ffb127
M
|
merge default
|
|
|||
| 2497 commits not shown. | ||||||