May 2017 Archives by thread
Starting: Mon May 1 00:32:38 EDT 2017
Ending: Wed May 31 18:18:38 EDT 2017
Messages: 363
- [pypy-commit] pypy cpyext-obj-stealing: fix tests for PyList_SetItem reference stealing
mattip
- [pypy-commit] pypy cpyext-obj-stealing: merge default into branch
mattip
- [pypy-commit] pypy default: Issue #1213: in ctypes, complain instead of silently ignoring _swappedbytes_
arigo
- [pypy-commit] pypy default: Expose win32's SetErrorMode, which is mandatory for the 3.5 test suite.
amauryfa at gmail.com
- [pypy-commit] pypy default: Backport _pypy_winbase_build.py from the 3.5 branch, to reduce merge conflicts.
amauryfa at gmail.com
- [pypy-commit] pypy py3.5: hg merge default
amauryfa
- [pypy-commit] stmgc c8-adaptive-trx-length: Merge number of segments in use function
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length: Make sequential transactions inevitable
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length: Fix name clash in macro expansion
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length: Implement logging of adaptive mode
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length: Merge all single thread mode enum values for consistency with analysis script
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length: Implement exponential build up and back off in transaction length
tobweber
- [pypy-commit] stmgc c8-overheads-instrumentation: Merge fix for name clash in macro expansion
tobweber
- [pypy-commit] pypy default: Add support for rejecting lone surrogates in utf16 and utf32 decoders.
rlamy
- [pypy-commit] pypy default: Attempt to fix build_cffi_imports on Windows: args should be passed as unicode strings.
rlamy
- [pypy-commit] pypy default: Popen.communicate() returns bytes if no encoding was specified.
rlamy
- [pypy-commit] pypy default: Attempt to import interp_scandir on Windows
rlamy
- [pypy-commit] pypy default: blindly change this from -1 to an arbitrary large number on 32 bit systems
rlamy
- [pypy-commit] pypy default: Backport changes from py3.5 branch
rlamy
- [pypy-commit] pypy default: Backport rposix changes from py3.5 branch
rlamy
- [pypy-commit] pypy default: Update includes README to mention new generated files, and the cpyext parse directory.
stefanor
- [pypy-commit] pypy PyBuffer-backport: Backport rpython.rlib.buffer changes from py3.5
rlamy
- [pypy-commit] pypy PyBuffer-backport: Backport memoryview changes
rlamy
- [pypy-commit] pypy py3.5: Expose msvcrt.SetErrorMode
amauryfa
- [pypy-commit] pypy py3.5: oops, SetErrorMode is in kernel32.dll
amauryfa
- [pypy-commit] pypy py3.5: Add a fake getwindowsversion()._platform_version, so that platform.win32_ver() can return something.
amauryfa
- [pypy-commit] stmgc c8-adaptive-trx-length-prolonged-backoff: Prolong the reduction in transaction length for a number of validations after a conflict
tobweber
- [pypy-commit] pypy faster-rstruct-2: start a branch in which to make rstruct faster in a more general way, by using llops.gc_load_indexed whenever it's possible, not only when reading from strings
antocuni
- [pypy-commit] pypy faster-rstruct-2: add a new Buffer.typed_read method, which does the same as the current strstorage.py: the idea is that this will be implemented by all Buffer subclasses for which it makes sense (e.g. ByteArrayBuffer can use the same gc_load_indexed approach as StringBuffer; and raw buffers should be able to use raw_load or similar). Copy the logic and the tests from strstorage.py, which will be deleted later
antocuni
- [pypy-commit] pypy faster-rstruct-2: add the compiled version of the tests. We still need to add the correct @specialize to Buffer.typed_read, although the tests pass anyway (because we always call Buffer.typed_read once)
antocuni
- [pypy-commit] pypy quad-color-gc: Update qcgc source files
ntruessel
- [pypy-commit] stmgc c8-adaptive-trx-length-prolonged-backoff: Use simple on off mechanism for the adaptive mode
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-prolonged-backoff: Backed out changeset 6b4b7aedc3d1
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-prolonged-backoff: Set exponential base of trx length modifier to 1000
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-prolonged-backoff: Set exponential base back to two because the effect was insignificant
tobweber
- [pypy-commit] pypy faster-rstruct-2: specialize on TP
antocuni
- [pypy-commit] pypy faster-rstruct-2: refactor rstruct.runpack to use the new Buffer.typed_read functionality, instead of deprecated strstorage. This breaks pypy/module/struct, whstruct, which will be fixed later
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix module/struct to use the new functionality. Add a passing test to check that struct.unpack(..., <str>) uses the fast-path; test_unpack_from still failing though
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement and test SubBuffer.typed_read; this fixes pypy/module/struct/test/test_struct.py:test_unpack_from
antocuni
- [pypy-commit] pypy faster-rstruct-2: kill as_str_and_offset_maybe, as it is no longer needed/used anywhere
antocuni
- [pypy-commit] pypy faster-rstruct-2: if we don't ensure that StringBuffer.value is not None, the annotator crashes as soon as you call str2chap(buf.as_str()). There are lots of places which do the assert before calling StringBuffer(), but I think it's simpler and better to just do it in the __init__, as it has no chances to work if value is None
antocuni
- [pypy-commit] pypy faster-rstruct-2: add docstring
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement typed_read for BytearrayBuffer and test that struct.unpack takes the fast path. However this is sub-optimal, see the comment for an explanation
antocuni
- [pypy-commit] pypy default: Populate tp_descr_get and tp_descr_set slots
amauryfa
- [pypy-commit] pypy default: Issue 2551: Struct should be initialized in __init__, not __new__
amauryfa
- [pypy-commit] pypy default: ast.parse(some_ast, ...) should return its first argument.
amauryfa
- [pypy-commit] pypy default: document that in f(**d) d must contain only string keys
cfbolz
- [pypy-commit] pypy default: Bytearrays have amortized constant-time "del a[:n]" (backport branch py3.5-bytearray)
rlamy
- [pypy-commit] pypy default: merge heads
rlamy
- [pypy-commit] pypy PyBuffer-backport: hg merge default
rlamy
- [pypy-commit] pypy PyBuffer-backport: Fix bytearray
rlamy
- [pypy-commit] pypy PyBuffer-backport: Add space.byte_w()
rlamy
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Implement a per thread adaptive single thread mode
tobweber
- [pypy-commit] pypy default: The issue is even more specific than documented
arigo
- [pypy-commit] pypy faster-rstruct-2: experimental checkin, reviews are welcome :). Refactor rgc._ResizableListSupportingRawPtr to use a proper lltype representation of its items, instead of a custom raw-malloced array. This makes it possible to implement ll_list(), i.e. the equivalent of llstr but for lists
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement rgc.ll_for_resizable_list, which is the equivalent of llstr(), but for lists
antocuni
- [pypy-commit] pypy default: Add another note
arigo
- [pypy-commit] pypy faster-rstruct-2: whoo, implement BytearrayBuffer.type_read using ll_for_resizable_list and llops.gc_load_index
antocuni
- [pypy-commit] pypy PyBuffer-backport: Backport cpyext changes
rlamy
- [pypy-commit] pypy PyBuffer-backport: _io and writebuf_w changes
rlamy
- [pypy-commit] pypy PyBuffer-backport: progress
rlamy
- [pypy-commit] pypy faster-rstruct-2: 1) avoid to take an unneeded string slice to skip the chars after typed_read; 2) as it was written before, we did the out-of-bound checks *after* the call to typed_read, which means that we risked to read past the end of the buffer
antocuni
- [pypy-commit] pypy default: update prebuilt external dependencies for win32
mattip
- [pypy-commit] pypy faster-rstruct-2: polish things a bit: rename rlist.make_LIST into LIST_OF, and use it everywhere instead of the ugly _ResizableListSupportingRawPtr._get_lltype()
antocuni
- [pypy-commit] cffi default: Issue 314
arigo
- [pypy-commit] cffi default: Try to systematically include the line number
arigo
- [pypy-commit] cffi default: Print the current directory (or the value tmpdir if explicit)
arigo
- [pypy-commit] pypy faster-rstruct-2: add rlib.buffer.RawBuffer, which automatically implements typed_read in terms of get_raw_address
antocuni
- [pypy-commit] pypy faster-rstruct-2: use the new RawBuffer as a base class for ArrayBuffer: as a consequence, now it automatically gets the benefits of the fastpath in rstruct :)
antocuni
- [pypy-commit] pypy faster-rstruct-2: use RawBuffer for all the various raw-memory backed buffers we have
antocuni
- [pypy-commit] cffi default: Py3 compat
arigo
- [pypy-commit] pypy faster-rstruct-2: rpython fix
antocuni
- [pypy-commit] pypy faster-rstruct-2: more rpython fix
antocuni
- [pypy-commit] pypy default: rename harmless copy-paste
mattip
- [pypy-commit] pypy default: fix table wrapping, from rackerlabs.github.io/docs-rackspace/tools/rtd-tables.html
mattip
- [pypy-commit] pypy default: back out b48484eba9d0, readthedocs is messed up
mattip
- [pypy-commit] pypy faster-rstruct-2: kill rlib/strstorage.py since it is no longer needed: the only place where it was still used was in the slow-path of float/double unpacking in runpck.py: replace it with the equivalent functionality exposed by StringBuffer.typed_read
antocuni
- [pypy-commit] pypy faster-rstruct-2: add rlib.Buffer.typed_read, and implement it for RawBuffer
antocuni
- [pypy-commit] pypy faster-rstruct-2: add MutableStringBuffer, which will be used by struct.pack to incrementally build the desired string
antocuni
- [pypy-commit] pypy default: win32: use jom.exe for parallel builds if available
mattip
- [pypy-commit] pypy str-measure: new branch
fijal
- [pypy-commit] pypy str-measure: a branch to start measuring how strings are used
fijal
- [pypy-commit] pypy str-measure: use a different id
fijal
- [pypy-commit] pypy PyBuffer-backport: Update micronumpy
rlamy
- [pypy-commit] pypy PyBuffer-backport: Add missing W_MemoryView.copy()
rlamy
- [pypy-commit] pypy faster-rstruct-2: add direct tests for the various pack() functions in rstruct
antocuni
- [pypy-commit] pypy faster-rstruct-2: add more tests
antocuni
- [pypy-commit] pypy faster-rstruct-2: WIP: break the world, and start to refactor pack_* to build the result inside a MutableStringBuffer instead of a StringBuilder
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix pack_int for the new interface
antocuni
- [pypy-commit] pypy faster-rstruct-2: migrate pack_float to the new buffer-based interface
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix pack_bool and pack_string
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix pack_pad
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix pack_pascal
antocuni
- [pypy-commit] pypy faster-rstruct-2: test&fix native's pack_{float, double}. In the meantime, reduce some code duplication between nativefmttable.py and ieee.py
antocuni
- [pypy-commit] pypy faster-rstruct-2: adapt pypy/module/struct to the new interface for packing, which means to use MutableStringBuffer instead of StrinbBuilder everywhere
antocuni
- [pypy-commit] pypy faster-rstruct-2: rpython fixes
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix marshal
antocuni
- [pypy-commit] pypy PyBuffer-backport: Restore py2 semantics for memoryview.__getitem__ and .__setitem__
rlamy
- [pypy-commit] pypy PyBuffer-backport: Ensure that buffer() returns read-only buffers
rlamy
- [pypy-commit] pypy PyBuffer-backport: cpyext fix
rlamy
- [pypy-commit] pypy faster-rstruct-2: I don't know why this popped out now as it seems unrelated with my recent changes: scoped_nonmovingbuffer expects data to be non-null, so we enforce the annotation to get an early error. Then, we fix the problem in interp_bz2.py
antocuni
- [pypy-commit] pypy default: Rewrite itertools.groupby(), following CPython instead of having many
arigo
- [pypy-commit] pypy py3.5: Manual merge of 6093ff1a44e6, plus reduce/setstate
arigo
- [pypy-commit] pypy faster-rstruct-2: we cannot malloc(STR, zero=True) because the GC does not support it. So, simplify a bit the code, remove the needs_zeros flag and always write zeros when it's needed
antocuni
- [pypy-commit] pypy faster-rstruct-2: remove the last remainder of the killed strstorage
antocuni
- [pypy-commit] pypy default: in theory, this is a fix
arigo
- [pypy-commit] pypy sthalik/fix-signed-integer-sizes-1494493539409: fix signed integer sizes
sthalik
- [pypy-commit] pypy default: Merged in sthalik/pypy-1/sthalik/fix-signed-integer-sizes-1494493539409 (pull request #543)
arigo
- [pypy-commit] pypy faster-rstruct-2: deleting strstorage tests was a mistake, because they were the only ones to test llop.gc_load_indexed. Rewrite them in a more direct way
antocuni
- [pypy-commit] pypy faster-rstruct-2: reduce code duplication
antocuni
- [pypy-commit] pypy faster-rstruct-2: add the JIT version of test_llop
antocuni
- [pypy-commit] pypy PyBuffer-backport: Fix memoryview.__setitem__, again
rlamy
- [pypy-commit] pypy PyBuffer-backport: fix _rawffi
rlamy
- [pypy-commit] pypy PyBuffer-backport: fix micronumpy
rlamy
- [pypy-commit] pypy PyBuffer-backport: Restrict marshalling to old-style buffers
rlamy
- [pypy-commit] pypy PyBuffer-backport: Don't bother preventing _numpypy.frombuffer() from working on memoryviews
rlamy
- [pypy-commit] pypy str-measure: improve the wrapper
fijal
- [pypy-commit] pypy default: Disable an assert that fails also on CPython, where failure is better just ignored at least on PyPy
arigo
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Use double instead of float
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Fix nested measurement of waiting time and time in validation;
tobweber
- [pypy-commit] pypy faster-rstruct-2: WIP: start to implement llop.gc_store_indexed; still missing implementation in the C backend and the JIT
antocuni
- [pypy-commit] pypy faster-rstruct-2: bah, actually TEST the rtyping of gc_store_index, and fix it
antocuni
- [pypy-commit] pypy faster-rstruct-2: test and implement gc_store_indexed in the C backend
antocuni
- [pypy-commit] pypy faster-rstruct-2: WIP: start to add support for llop.gc_store_indexed in the JIT, which means to add stuff a bit everywhere. But interp_operations does not support returning GCREFs, so we need to tweak the tests differently
antocuni
- [pypy-commit] pypy faster-rstruct-2: hoorray! Implement the last bits of gc_store_indexed in llgraph and finally the test passes :)
antocuni
- [pypy-commit] pypy faster-rstruct-2: add JIT support for gc_store_indexed of floats
antocuni
- [pypy-commit] pypy faster-rstruct-2: add a passing test for single floats
antocuni
- [pypy-commit] pypy faster-rstruct-2: shuffle the order of arguments of llop.gc_store_indexed to match the existing rop.GC_STORE_INDEXED
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement support for gc_store_indexed also in llsupport: this fixes the tests for the x86 backend, and hopefully for the other backends as well
antocuni
- [pypy-commit] pypy faster-rstruct-2: add the llop test also for ARM
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement MutableStringBuffer.typed_write
antocuni
- [pypy-commit] pypy faster-rstruct-2: finally! Add a fastpath for packing ints :)
antocuni
- [pypy-commit] pypy PyBuffer-backport: Allow memoryviews in compile() and _codecs.unicode_internal_decode()
rlamy
- [pypy-commit] pypy default: Document merged branch
rlamy
- [pypy-commit] pypy faster-rstruct-2: fix bhimpl_gc_store_indexed_i, which was not tested because the blackhole didn't see the op :(
antocuni
- [pypy-commit] pypy faster-rstruct-2: unroll the loop if count is a small constant
antocuni
- [pypy-commit] pypy py3.5: fix translation
rlamy
- [pypy-commit] pypy py3.5: Implement PyModule_GetState()
arigo
- [pypy-commit] pypy default: issue #2557: file.read(1) could return 2 bytes on Windows
arigo
- [pypy-commit] pypy faster-rstruct-2: add a fastpath for floats and doubles
antocuni
- [pypy-commit] pypy faster-rstruct-2: use the fast path also for the native float/double packing
antocuni
- [pypy-commit] pypy faster-rstruct-2: rename fmtiter.result into wbuf, which makes more sense
antocuni
- [pypy-commit] pypy faster-rstruct-2: pass an external wbuf to PackFormatIterator
antocuni
- [pypy-commit] pypy faster-rstruct-2: now that we have all the necessary infrastructure, implement pack_into in a more efficient way, so that PackFormatIterator writes directly inside the destination buffer (using the fast paths if possible)
antocuni
- [pypy-commit] pypy faster-rstruct-2: whoo... finally reach the whole point of the branch: struct.pack_into(bytearray...) takes the fast path :)
antocuni
- [pypy-commit] pypy faster-rstruct-2: we cannot use the fastpath for '<f' or '>f', because we need to check for overflow in that case. It is fine to use the fastpath for doubles though, because there is no risk of overflow
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix the tests in jit/backend/x86/test/test_llop.py:test_gc_store*
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix the test_pypy_c tests about the struct module, now that we have the fast path also for packing
antocuni
- [pypy-commit] pypy faster-rstruct-2: add tests for the fast paths of struct.unpack on raw buffers and bytearrays
antocuni
- [pypy-commit] pypy faster-rstruct-2: add tests for the fast path of pack_into
antocuni
- [pypy-commit] pypy default: OpenBSD also needs sys/ttycom.h included.
ltratt
- [pypy-commit] pypy default: string.h needs to be included for strlen to be found.
ltratt
- [pypy-commit] pypy default: Disable vmprof on OpenBSD as it doesn't build.
ltratt
- [pypy-commit] pypy default: hg merge default
ltratt
- [pypy-commit] pypy faster-rstruct-2: WIP: hg merge default: lots of conflicts due to the merge of Pybuffer-backport: textually fixed the conflicts but did not run the tests yet
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix tests
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix BytearrayBuffer.typed_{read, write}
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix for the recent changes to Bytearray, which has got an extra _offset field now. This code is a bit suboptimal because it contains lots of guards :(
antocuni
- [pypy-commit] pypy faster-rstruct-2: the Bytebuffer class in rlib.buffer and pypy.module.__pypy__ was almost identical. Kill the latter
antocuni
- [pypy-commit] pypy faster-rstruct-2: move the definition of the class after RawBuffer, which is a base class and thus is better positioned near the top of the file
antocuni
- [pypy-commit] pypy faster-rstruct-2: introduce GCBuffer, to share the implementation of typed_read and type_write among all buffers which are baked by GC-managed memory. Use it for StringBuffer and BytearrayBuffer
antocuni
- [pypy-commit] pypy faster-rstruct-2: write tests for ByteBuffer, make it a subclass of GCBuffer and reduce a bit of code duplication with ByteArrayBuffer
antocuni
- [pypy-commit] pypy faster-rstruct-2: bah
antocuni
- [pypy-commit] pypy faster-rstruct-2: rpython fix, else the types returned by _get_gc_data conflict
antocuni
- [pypy-commit] pypy faster-rstruct-2: bah, the base_ofs needs to be proven constant at translation time. Try to help the rtyper
antocuni
- [pypy-commit] pypy faster-rstruct-2: rewrite buffer.GCBuffer as a class decorator instead of a base class, because we need to provide different, specialized versions of typed_{read, write} in which base_ofs is a constant to make the codewriter happy
antocuni
- [pypy-commit] pypy faster-rstruct-2: use @GCBuffer to implement MutableStringBuffer
antocuni
- [pypy-commit] pypy default: add test that passes with -A
mattip
- [pypy-commit] pypy default: add slot definitions for __len__, __setitem__, __delitem__, avoid crash in PyDict_Next
mattip
- [pypy-commit] pypy default: fix test
mattip
- [pypy-commit] pypy py3.5: Use correct keyword argument for strftime() routines in datetime
danchr
- [pypy-commit] pypy cpyext-recursionlimit: add test, implement recursion limit functions with ugly hack for untranslated test
mattip
- [pypy-commit] pypy py3.5: Call sys.__interactivehook__ (issue #2558)
rlamy
- [pypy-commit] pypy py3.5: Document PyBuffer
rlamy
- [pypy-commit] pypy py3.5: fix test
rlamy
- [pypy-commit] pypy cpyext-recursionlimit: fix translation
mattip
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Remove former transaction length back off mechanism and comment out debug output
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Remove min transaction length (is now zero)
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Initialize single thread mode properties of thread local in setup.c
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Do not set transactions inevitable on start
tobweber
- [pypy-commit] stmgc c8-adaptive-trx-length-per-thread: Add timing macros to hide internals from user code when generating custom payload events
tobweber
- [pypy-commit] pypy faster-rstruct-2: move setzeros into the base class rlib.buffer.Buffer, because it is used by rstruct to pack and needs to be available on all buffers
antocuni
- [pypy-commit] pypy faster-rstruct-2: WIP: add a failing test, which is triggered when W_BytearrayObject._offset > 0
antocuni
- [pypy-commit] pypy faster-rstruct-2: turn GCBuffer into a base class again :(, so that we will be able to provide default methods
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix test_pack_into_bytearray_delete
antocuni
- [pypy-commit] pypy py3.5: Run sys.__interactivehook__() also when using -i
rlamy
- [pypy-commit] pypy py3.5: Disable enablerlcompleter() since it's broken on PyPy (note that virtualenv's site.py doesn't use it either)
rlamy
- [pypy-commit] pypy py3.5: Reenable the parts of enablerlcompleter() that do work (thanks o11c!)
rlamy
- [pypy-commit] pypy default: tweak for cython
mattip
- [pypy-commit] pypy default: a readme for the applevel tests in this dir
cfbolz
- [pypy-commit] pypy faster-rstruct-2: ouch, this was plainly wrong, as it overwrote the correct teardown_method just above
antocuni
- [pypy-commit] pypy faster-rstruct-2: write tests to check that USE_FASTPATH and ALLOW_SLOWPATH actually do what they are supposed to do
antocuni
- [pypy-commit] pypy faster-rstruct-2: add a new test flag to disallow fast paths
antocuni
- [pypy-commit] pypy faster-rstruct-2: add a passing test to check that we don't take the fast path for unaligned access
antocuni
- [pypy-commit] pypy faster-rstruct-2: WIP: add a failing test which show why the current way to detect alignement is buggy
antocuni
- [pypy-commit] pypy faster-rstruct-2: delegate the alignment check to Buffer.typed_{read, write}, to take in account also the SubBuffer's offset. Moreover, reuse the logic in rawstorage to determine whether it is fine to do an unaligned access, depending on the CPU
antocuni
- [pypy-commit] pypy faster-rstruct-2: hg merge default
antocuni
- [pypy-commit] pypy.org extradoc: OS/X => OS X
arigo
- [pypy-commit] pypy py3.5: Work around issue #2560
rlamy
- [pypy-commit] pypy py3.5: Fix PY_VERSION inconsistency
rlamy
- [pypy-commit] pypy faster-rstruct-2: rpython fix for 32 bit
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix the expected JIT ops after the recent W_BytearrayObject changes
antocuni
- [pypy-commit] pypy pypy_ctypes_nosegfault_nofastpath: fixed ctypes segfault by removing fastpath. Test in test_segfault.py
Dodan
- [pypy-commit] pypy pypy_ctypes_nosegfault_nofastpath: Test_segfault.py now throws ctypes.ArgumentError
Dodan
- [pypy-commit] pypy pypy_ctypes_nosegfault_nofastpath: Removed test_segfault.py. The test was in test_functions.py in test_argument_conversion_and_checks
Dodan
- [pypy-commit] pypy default: Merged in Dodan/pypy_ctypes/pypy_ctypes_nosegfault_nofastpath (pull request #547)
cfbolz
- [pypy-commit] pypy faster-rstruct-2: bah, the two tests shared the same cache, which means that they worked if invoked separately but broke if you tested the whole file
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix writeanalyze.py to take into account llop.gc_store_indexed
antocuni
- [pypy-commit] pypy cpyext-obj-stealing: Tweaks and simplifications
arigo
- [pypy-commit] pypy cpyext-unhashable: test, fix for unhashable c-api objects
mattip
- [pypy-commit] pypy default: merge cpyext-obj-stealing which improves PyListObject refcounting compatibility
mattip
- [pypy-commit] pypy default: merge cpyext-recursionlimit which implements Py_EnterRecursiveCall and friends
mattip
- [pypy-commit] pypy default: document merged branches
mattip
- [pypy-commit] pypy default: Remove unused translation option 'withstrbuf' and supporting code
rlamy
- [pypy-commit] pypy strbufobject: A branch to play with W_StringBufferObject (recently deleted on default)
rlamy
- [pypy-commit] pypy strbufobject: backout c74e33eeef0d: undelete W_StringBufferObject and withstrbuf config option
rlamy
- [pypy-commit] pypy default: fix translation
mattip
- [pypy-commit] pypy faster-rstruct-2: fix a misuse of box.getfloat() vs box.getfloatstorage(): this was harmless on 64 bit of course, but caused test_llop:test_gc_store_indexed_double to fail on 32 bit
antocuni
- [pypy-commit] pypy cpyext-unhashable: merge default into branch
mattip
- [pypy-commit] pypy cpyext-unhashable: test, fix for cpytext hash of empty string subclass not returning -1
mattip
- [pypy-commit] pypy faster-rstruct-2: enforce the annotation of write_float_at_mem: the problem fixed by 0e91701e92f0 was that we mixed a r_floatstorage with a float: on 32 bit they are different types but the annotator union()ed them to Float, causing nonsense at every place calling write_float_at_mem. The goal of this commit is to catch this kind of bug at translation time
antocuni
- [pypy-commit] pypy default: Build main binaries on OpenBSD with wxallowed.
ltratt
- [pypy-commit] pypy default: Merge default
ltratt
- [pypy-commit] pypy default: fix issue #1877: replace bare Exceptions with AnnotatorError in the annotator
cfbolz
- [pypy-commit] pypy sockopt_zero: added the possibility to have zero sized buffer length
Alecsandru Patrascu
- [pypy-commit] pypy sockopt_zero: proper fix after review
Alecsandru Patrascu
- [pypy-commit] pypy sockopt_zero: shorten code path
Alecsandru Patrascu
- [pypy-commit] pypy default: Merged in palecsandru/pypy2_sockopt_zero/sockopt_zero (pull request #548)
cfbolz
- [pypy-commit] pypy py3.5: Adapt test to py3
rlamy
- [pypy-commit] pypy cpyext-unhashable: add passing tests
mattip
- [pypy-commit] pypy py3.5: Add validation of __getnewargs__()/__getnewargs_ex__() return values
rlamy
- [pypy-commit] pypy default: Backout PR #490 (ab8a0b092d2d + follow-up bd7b1902c4df).
arigo
- [pypy-commit] pypy faster-rstruct-2: backout 85d3ab6fe80b, which was broken, and start to rewrite the logic in a more generic way. Also, write tests in such a way that we can check that gc_{load, store}_indexed produce effects corresponding to their equivalent
antocuni
- [pypy-commit] pypy faster-rstruct-2: add support & test for gc_load_indexed of a list of chars
antocuni
- [pypy-commit] pypy faster-rstruct-2: implement the analyzer for gc_store_indexed
antocuni
- [pypy-commit] pypy faster-rstruct-2: add a passing test for gc_store_indexed on a list of chars
antocuni
- [pypy-commit] pypy cpyext-unhashable: backed out 552622605578, breaks too many things
mattip
- [pypy-commit] pypy faster-rstruct-2: @enforceargs does not play well with @specialize, use _annenforceargs_ directly
antocuni
- [pypy-commit] pypy faster-rstruct-2: typo
antocuni
- [pypy-commit] pypy faster-rstruct-2: fix test_gc_load_indexed_str
antocuni
- [pypy-commit] pypy faster-rstruct-2: this this test_pypy_c test: now struct.unpack generates a fast raw_load_i; use 'i' instead of '<i' to make sure that we get the same code even on big endian machines
antocuni
- [pypy-commit] pypy faster-rstruct-2: make this test endianess-independent: hopefully this fixed the failure on S390X
antocuni
- [pypy-commit] pypy default: fix test_whatsnew
antocuni
- [pypy-commit] pypy default: kill this test: originally it was meant to check that we take the fastpath when doing a ctypes call, but now the fastpath has gone and thus there is nothing to test. Moreover, most of the logic has been skipped since forever, apparently
antocuni
- [pypy-commit] pypy better-test-whatsnew: a branch where to improve test_whatsnew:
antocuni
- [pypy-commit] pypy better-test-whatsnew: improve this test and make sure that the CURRENT branch is documented: this way we get a failure early before merging the branch instead of having test_whatsnew failing on default the day after
antocuni
- [pypy-commit] pypy better-test-whatsnew: add this branch to the documented ones, and make the test passing again
antocuni
- [pypy-commit] pypy better-test-whatsnew: close branch about to be merged
antocuni
- [pypy-commit] pypy default: merge the branch better-test-whatsnew: now you must document your branch in whatsnew-HEAD.rst BEFORE merging it
antocuni
- [pypy-commit] pypy faster-rstruct-2: remove dead import (thanks cfbolz)
antocuni
- [pypy-commit] pypy Kounavi/fix-typo-depricate-to-deprecate-p-1495624547235: Fix typo (depricate to deprecate) :P
Kounavi
- [pypy-commit] pypy default: Merged in Kounavi/pypy/Kounavi/fix-typo-depricate-to-deprecate-p-1495624547235 (pull request #550)
arigo
- [pypy-commit] pypy faster-rstruct-2: close branch about to be merged
antocuni
- [pypy-commit] pypy default: merge the faster-rstruct-2 branch: this makes struct.pack and struct.pack_into much faster by using raw_store or gc_store_indexed whenever possible. Moreover, it enables the existing struct.unpack fast path to all the existing buffer types, whereas previously it was enabled only for strings
antocuni
- [pypy-commit] pypy default: document the merged branch
antocuni
- [pypy-commit] pypy PyPy_profopt_enabled: PyPy 5.8.0 functional profopt. Precleanup
Dodan
- [pypy-commit] pypy PyPy_profopt_enabled: PyPy-5.8.0-Profopt-enabled-postclean
Dodan
- [pypy-commit] pypy PyPy_profopt_enabled: Profopt option now can come paired with --profoptpath to specify the training workload
Dodan
- [pypy-commit] pypy py3.5: Everything in lib_pypy should at least be compatible with py3k syntax
stefanor
- [pypy-commit] pypy default: don't test the blackhole path in ARM tests, for the very same reason as we do in x86 ones
antocuni
- [pypy-commit] pypy default: Fix space.bufferstr_w() returning a wrong value for numpy arrays and add an indirect test for this
rlamy
- [pypy-commit] pypy py3.5: __wbuffer__ and __rbuffer__ aren't used in pypy3
rlamy
- [pypy-commit] pypy py3.5: update test with current pypy error messages
rlamy
- [pypy-commit] pypy default: No clue how this missing @specialize was not noticed earlier.
arigo
- [pypy-commit] pypy cpyext-unhashable: more places to ensure cpyext hashing an empty string-like does not return -1
mattip
- [pypy-commit] pypy default: more places to convert hash value -1 to -2, see comments to issue #2346
mattip
- [pypy-commit] pypy default: do not go through descroperation.binop_impl again, call descr_add directly
mattip
- [pypy-commit] pypy cpyext-unhashable: close exploratory branch
mattip
- [pypy-commit] pypy default: Fix a rare stacklet issue that was there since a while with shadowstack.
arigo
- [pypy-commit] pypy default: Fix for PR #540 on non-Linux platforms
arigo
- [pypy-commit] pypy default: Add a marker into the shadowstack from the JIT. Might improve performance in some cases
arigo
- [pypy-commit] pypy py3.5: Rename ArrayData to ArrayBuffer
rlamy
- [pypy-commit] pypy default: Reduce diff with py3.5
rlamy
- [pypy-commit] pypy default: Argh, fix test that was not testing anything
arigo
- [pypy-commit] pypy default: Performance tweaks to round(x, n) for the case n == 0
arigo
- [pypy-commit] pypy default: Skip the test about byte-swapped structures, for now
arigo
- [pypy-commit] pypy default: Fix for 0e27a73c87f6
arigo
- [pypy-commit] pypy default: Make sure sys.prefix is always defined, even if it contains the
arigo
- [pypy-commit] pypy default: improve zipfile performance by not doing repeated string concatenation
planter
- [pypy-commit] pypy default: Add checks on the length argument given to socket.getsockopt(), like CPython
arigo
- [pypy-commit] cffi default: tweaks
arigo
- [pypy-commit] cffi sirtom67/float_complex: hg merge default
arigo
- [pypy-commit] cffi sirtom67/float_complex: Minor style clean-ups. Update the test to conform more closely to the
arigo
- [pypy-commit] cffi sirtom67/float_complex: Fix test, fix all abort()
arigo
- [pypy-commit] cffi sirtom67/float_complex: Test explicitly that if we try to call a function in ABI mode with a
arigo
- [pypy-commit] cffi sirtom67/float_complex: Tweaks
arigo
- [pypy-commit] cffi sirtom67/float_complex: MSVC might not support _Complex (and the include <complex.h>).
arigo
- [pypy-commit] cffi sirtom67/float_complex: comment
arigo
- [pypy-commit] cffi sirtom67/float_complex: Revert the changes to vengine_cpy: better to say that the
arigo
- [pypy-commit] cffi sirtom67/float_complex: Support parsing "float _Complex" and "double _Complex" inside parse_c_type
arigo
- [pypy-commit] cffi sirtom67/float_complex: More tweaks, more tests, try harder to avoid including <complex.h>
arigo
- [pypy-commit] cffi sirtom67/float_complex: fix
arigo
- [pypy-commit] cffi sirtom67/float_complex: MSVC fixes
arigo
- [pypy-commit] pypy default: test, add a failure path, and document differences in PyTuple_SetItem
mattip
- [pypy-commit] pypy default: tweak documentation
mattip
- [pypy-commit] pypy default: check tuple, not contents
mattip
- [pypy-commit] pypy default: expand test, fix translation
mattip
- [pypy-commit] pypy jumbojet: sched_get min/max
nanjekye
- [pypy-commit] pypy py3.5: hg merge jumbojet
arigo
- [pypy-commit] pypy default: merge back changes to RPython in py3.5 (3ee9941e823c)
arigo
- [pypy-commit] pypy py3.5: typo
cfbolz
- [pypy-commit] pypy py3.5: some missing docstrings
cfbolz
- [pypy-commit] pypy release-pypy2.7-5.x: merge default into release
mattip
- [pypy-commit] pypy release-pypy2.7-5.x: update version to 5.8.0 final
mattip
- [pypy-commit] pypy release-pypy3.5-5.x: merge py3.5 into release branch
mattip
- [pypy-commit] pypy release-pypy3.5-5.x: update version to 5.8.0 beta
mattip
- [pypy-commit] pypy py3.5: update version to 5.9.0, restart whatsnew part 1
mattip
- [pypy-commit] pypy py3.5: restart whatsnew part 2
mattip
- [pypy-commit] pypy default: update development version to 5.9, restart whatsnew (part 1)
mattip
- [pypy-commit] pypy default: start very rough draft of release notice, restart whatsnew (part 2)
mattip
- [pypy-commit] pypy release-pypy2.7-5.x: merge default into release branch
mattip
- [pypy-commit] pypy py3.5: fixes for whatsnew, document merged branch
mattip
- [pypy-commit] pypy default: add this to default too
mattip
- [pypy-commit] pypy default: fix first commit after release
mattip
- [pypy-commit] pypy default: start to summarize changes since 5.7.0, read up to changeset 198dc138680f
mattip
- [pypy-commit] cffi sirtom67/float_complex: Documentation
arigo
- [pypy-commit] cffi sirtom67/float_complex: Close branch, ready to merge
arigo
- [pypy-commit] cffi default: Merge branch sirtom67/float_complex, adding support for complex numbers.
arigo
- [pypy-commit] cffi default: pypy tweak
arigo
- [pypy-commit] pypy cffi-complex: A branch to implement complex numbers from cffi 1.11
arigo
- [pypy-commit] cffi default: Whatsnew
arigo
- [pypy-commit] cffi char16_char32_t: A branch for issue #315
arigo
- [pypy-commit] pypy rawrefcount-review: move code
rlamy
- [pypy-commit] pypy rawrefcount-review: Remove some code duplication
rlamy
- [pypy-commit] cffi char16_char32_t: fixes
arigo
- [pypy-commit] cffi char16_char32_t: Tests and fixes
arigo
- [pypy-commit] cffi char16_char32_t: more tests
arigo
- [pypy-commit] cffi char16_char32_t: in-progress
arigo
- [pypy-commit] cffi char16_char32_t: tweaks
arigo
- [pypy-commit] pypy cffi-complex: Implement complexes, enough to make the test_c part pass
arigo
- [pypy-commit] pypy cffi-complex: parse_c_type
arigo
- [pypy-commit] cffi default: no-op, but avoids confusion because these two are always listed in the
arigo
- [pypy-commit] cffi default: Update the version to 1.11
arigo
- [pypy-commit] pypy cffi-complex: Call support for functions with complex args/return types
arigo
- [pypy-commit] pypy cffi-complex: Update the version number here
arigo
- [pypy-commit] cffi default: update
arigo
- [pypy-commit] pypy cffi-complex: update
arigo
- [pypy-commit] pypy py3.5: Update _testcapimodule.c to 3.5.3, disabling things that don't compile
rlamy
- [pypy-commit] pypy py3.5: Skip crashing test
rlamy
- [pypy-commit] pypy cffi-complex: import cffi/e4e863b06bc7
arigo
- [pypy-commit] pypy cffi-complex: Document branch
arigo
- [pypy-commit] pypy cffi-complex: ready to merge
arigo
- [pypy-commit] pypy default: hg merge cffi-complex
arigo
- [pypy-commit] pypy default: summarize changes since 5.7.0, read up to changeset 77bf5cbc29e9
mattip
Last message date:
Wed May 31 18:18:38 EDT 2017
Archived on: Wed May 31 18:18:40 EDT 2017
This archive was generated by
Pipermail 0.09 (Mailman edition).