January 2016 Archives by subject
Starting: Fri Jan 1 03:32:33 EST 2016
Ending: Sun Jan 31 18:42:22 EST 2016
Messages: 660
- [pypy-commit] benchmarks default: add dropbox/pyxl benchmark
fijal
- [pypy-commit] benchmarks default: add some other pyston benchmarks
fijal
- [pypy-commit] buildbot default: HACK: don't upload branch names any more. see discussion on #pypy.
arigo
- [pypy-commit] buildbot default: merged s390x changes
plan_rich
- [pypy-commit] buildbot default: renamed ibm-research to dje, added seperate nightly to test not default on s390x, but s390x-backend
plan_rich
- [pypy-commit] buildbot s390x-buildbot: added more config:
plan_rich
- [pypy-commit] buildbot s390x-buildbot: added s390x to config
plan_rich
- [pypy-commit] buildbot s390x-buildbot: close branch
plan_rich
- [pypy-commit] buildbot s390x-buildbot: include linux in the value of the platform dict
plan_rich
- [pypy-commit] cffi default: Add a passing test about "from mymodule import *"
arigo
- [pypy-commit] cffi default: Add vararg demo
arigo
- [pypy-commit] cffi default: Allow testing against installed package (which has egg_info already)
stefanor
- [pypy-commit] cffi default: Bump version number here
arigo
- [pypy-commit] cffi default: bump version number to 1.5.0
arigo
- [pypy-commit] cffi default: clarify (thanks Rémi)
arigo
- [pypy-commit] cffi default: Delay changing this to 1.6 (1.5 is very soon after 1.4)
arigo
- [pypy-commit] cffi default: Don't blow away PYTHONPATH
stefanor
- [pypy-commit] cffi default: Don't hardcode _d. If that is the extension, it'll be in sysconfig
stefanor
- [pypy-commit] cffi default: emphasis
arigo
- [pypy-commit] cffi default: Expand the error message
arigo
- [pypy-commit] cffi default: fix
arigo
- [pypy-commit] cffi default: fix
arigo
- [pypy-commit] cffi default: fixes for pypy
arigo
- [pypy-commit] cffi default: fix when running this test in the same process as ../cffi0/test_verify.py (found on os/x, but I don't know why it doesn't also fail on linux)
arigo
- [pypy-commit] cffi default: here too
arigo
- [pypy-commit] cffi default: hg merge demo-cleanup
arigo
- [pypy-commit] cffi default: hg merge release-1.5
arigo
- [pypy-commit] cffi default: hg merge static-callback-embedding
arigo
- [pypy-commit] cffi default: hg merge windows-tls
arigo
- [pypy-commit] cffi default: issue #243: explicitly complain if ffi.def_extern() is called on the
arigo
- [pypy-commit] cffi default: Issue #244: parse_type() calls the Parser logic but self._options used
arigo
- [pypy-commit] cffi default: Missing .c and .h files
stefanor
- [pypy-commit] cffi default: more tests pass
arigo
- [pypy-commit] cffi default: OS/X: in-progress, one test now passes
arigo
- [pypy-commit] cffi default: OS/X fix: clang has got a warning enabled by default here
arigo
- [pypy-commit] cffi default: Print all commands executed with all details
arigo
- [pypy-commit] cffi default: Support extensions for pydebug cpythons
stefanor
- [pypy-commit] cffi default: Trying a different hack: stop patching SO and EXT_SUFFIX in
arigo
- [pypy-commit] cffi default: Use mutex/condition variables instead of semaphores (for os/x)
arigo
- [pypy-commit] cffi default: Use os.pathsep instead of directly ':'
arigo
- [pypy-commit] cffi default: What's New
arigo
- [pypy-commit] cffi default: win32
arigo
- [pypy-commit] cffi default: win32: needs /MANIFEST
arigo
- [pypy-commit] cffi default: Windows uses '_d' even though I'm pretty sure DEBUG_EXT is a Debian-only
arigo
- [pypy-commit] cffi demo-cleanup: fix the demo
arigo
- [pypy-commit] cffi demo-cleanup: just killing the verify() line is not enough
arigo
- [pypy-commit] cffi demo-cleanup: ready to merge
arigo
- [pypy-commit] cffi demo-cleanup: tweaks
arigo
- [pypy-commit] cffi demo-cleanup: tweaks
arigo
- [pypy-commit] cffi demo-cleanup: tweaks
arigo
- [pypy-commit] cffi demo-cleanup: tweaks
arigo
- [pypy-commit] cffi demo-cleanup: tweaks
arigo
- [pypy-commit] cffi embedding-pypy-win32: allow embedding tests to run on pypy win32
mattip
- [pypy-commit] cffi release-1.5: make release branch
arigo
- [pypy-commit] cffi release-1.5: md5/sha1
arigo
- [pypy-commit] cffi static-callback-embedding: (untested) trying to have the multithreaded tests run on windows
arigo
- [pypy-commit] cffi static-callback-embedding: Add a simple test (cpython 2.7 only)
arigo
- [pypy-commit] cffi static-callback-embedding: Add some sections
arigo
- [pypy-commit] cffi static-callback-embedding: Allow platform-specific hacks to invoke the compiler
arigo
- [pypy-commit] cffi static-callback-embedding: An apparently different way of failing
arigo
- [pypy-commit] cffi static-callback-embedding: A test checking that thread-local values are saved, even though
arigo
- [pypy-commit] cffi static-callback-embedding: A test of loading different cffi embedded modules in different threads.
arigo
- [pypy-commit] cffi static-callback-embedding: A test with two modules.
arigo
- [pypy-commit] cffi static-callback-embedding: Avoid setting LD_LIBRARY_PATH
arigo
- [pypy-commit] cffi static-callback-embedding: Bah
arigo
- [pypy-commit] cffi static-callback-embedding: Check that the module is present in sys.modules at this point,
arigo
- [pypy-commit] cffi static-callback-embedding: details
arigo
- [pypy-commit] cffi static-callback-embedding: Finalize and review embedding.rst.
arigo
- [pypy-commit] cffi static-callback-embedding: fix
arigo
- [pypy-commit] cffi static-callback-embedding: fix
arigo
- [pypy-commit] cffi static-callback-embedding: fix
arigo
- [pypy-commit] cffi static-callback-embedding: fix docstring
arigo
- [pypy-commit] cffi static-callback-embedding: fixes
arigo
- [pypy-commit] cffi static-callback-embedding: Fix ffi.compile() to automatically link with the python library
arigo
- [pypy-commit] cffi static-callback-embedding: Fix some comments by fijal
arigo
- [pypy-commit] cffi static-callback-embedding: Fix the multithreaded initialization.
arigo
- [pypy-commit] cffi static-callback-embedding: Found out how to control the extension of distutils-built libraries.
arigo
- [pypy-commit] cffi static-callback-embedding: Generalize a bit
arigo
- [pypy-commit] cffi static-callback-embedding: Give up testing windows: it kinda seems to work but I really, really,
arigo
- [pypy-commit] cffi static-callback-embedding: hg merge default
arigo
- [pypy-commit] cffi static-callback-embedding: improve test, but it is failing under windows
arigo
- [pypy-commit] cffi static-callback-embedding: in-progress: a comment about eggs, and hacks to run the tests on pypy (some failures so far)
arigo
- [pypy-commit] cffi static-callback-embedding: In-progress: fix tests to attempt to use distutils also for compiling
arigo
- [pypy-commit] cffi static-callback-embedding: in-progress: work work work to port this to Python 3
arigo
- [pypy-commit] cffi static-callback-embedding: kill empty lines at the start and dedent manually, to give
arigo
- [pypy-commit] cffi static-callback-embedding: Maybe it's clearer this way, with an API that matches the intent rather
arigo
- [pypy-commit] cffi static-callback-embedding: merge heads
arigo
- [pypy-commit] cffi static-callback-embedding: minor stuff
arigo
- [pypy-commit] cffi static-callback-embedding: more doc
arigo
- [pypy-commit] cffi static-callback-embedding: more doc
arigo
- [pypy-commit] cffi static-callback-embedding: More Windows hacks
arigo
- [pypy-commit] cffi static-callback-embedding: One more test, passing
arigo
- [pypy-commit] cffi static-callback-embedding: possibly clarify example by using the syntax that allows naturally
arigo
- [pypy-commit] cffi static-callback-embedding: Pseudo-tests that print some performance numbers for calling an embedded
arigo
- [pypy-commit] cffi static-callback-embedding: py3 compat
arigo
- [pypy-commit] cffi static-callback-embedding: Py3 fix
arigo
- [pypy-commit] cffi static-callback-embedding: Py3 fix
arigo
- [pypy-commit] cffi static-callback-embedding: Py3 fixes. Skip the embedding tests if linking with -lpython%d.%d
arigo
- [pypy-commit] cffi static-callback-embedding: ready to merge
arigo
- [pypy-commit] cffi static-callback-embedding: refactor details, start writing docs
arigo
- [pypy-commit] cffi static-callback-embedding: Skip this test on windows
arigo
- [pypy-commit] cffi static-callback-embedding: Stop running the embedding_init_code() code as if it was part of the
arigo
- [pypy-commit] cffi static-callback-embedding: Test and fix
arigo
- [pypy-commit] cffi static-callback-embedding: test for not invoking the init code recursively
arigo
- [pypy-commit] cffi static-callback-embedding: test multiple threads all doing the initial call to an "extern Python"
arigo
- [pypy-commit] cffi static-callback-embedding: tried hard and failed to have the embedding tests run in a debug-mode python
arigo
- [pypy-commit] cffi static-callback-embedding: try
arigo
- [pypy-commit] cffi static-callback-embedding: try
arigo
- [pypy-commit] cffi static-callback-embedding: trying win fix
arigo
- [pypy-commit] cffi static-callback-embedding: two tests pass on windows! progress
arigo
- [pypy-commit] cffi static-callback-embedding: updates
arigo
- [pypy-commit] cffi static-callback-embedding: Update the version numbers and make sure the _cffi_backend module is
arigo
- [pypy-commit] cffi static-callback-embedding: Ups, found the bug
arigo
- [pypy-commit] cffi static-callback-embedding: use declspec(dllimport/dllexport) on windows
arigo
- [pypy-commit] cffi static-callback-embedding: We need fflush(stdout) here, for a possibly bad reason: cpython 2.x
arigo
- [pypy-commit] cffi static-callback-embedding: windows fix
arigo
- [pypy-commit] cffi static-callback-embedding: Write a line when we run the final executable too
arigo
- [pypy-commit] cffi windows-tls: a branch to improve the TLS handling on Windows (and fix a small leak)
arigo
- [pypy-commit] cffi windows-tls: fix
arigo
- [pypy-commit] cffi windows-tls: fixes
arigo
- [pypy-commit] cffi windows-tls: Generalize the code for 'local_thread_state' and move it to
arigo
- [pypy-commit] cffi windows-tls: ready to merge
arigo
- [pypy-commit] cffi windows-tls: Remove this field if it's not used
arigo
- [pypy-commit] cffi windows-tls: trying out with a DllMain function
arigo
- [pypy-commit] extradoc extradoc: add myself
fijal
- [pypy-commit] extradoc extradoc: add myself, departing Thurs evening to catch a Fri morning flight from Zurich
mattip
- [pypy-commit] extradoc extradoc: add myself to list of people
cfbolz
- [pypy-commit] extradoc extradoc: Add myself to the list of people coming to the Leysin sprint.
mjacob
- [pypy-commit] extradoc extradoc: Clarify?
arigo
- [pypy-commit] extradoc extradoc: Draft blog post
arigo
- [pypy-commit] extradoc extradoc: fix merge
fijal
- [pypy-commit] extradoc extradoc: Fix year.
mjacob
- [pypy-commit] extradoc extradoc: I'm coming
rlamy
- [pypy-commit] extradoc extradoc: I'm comming too.
plan_rich
- [pypy-commit] extradoc extradoc: in-progress
arigo
- [pypy-commit] extradoc extradoc: Leysin Winter Sprint 2016
arigo
- [pypy-commit] extradoc extradoc: me -> to the people.txt list for the leysin sprint
plan_rich
- [pypy-commit] extradoc extradoc: precision
arigo
- [pypy-commit] extradoc extradoc: register for Leysin sprint
Raemi
- [pypy-commit] extradoc extradoc: Registering for the Winter Sprint
Pierre-Yves David
- [pypy-commit] extradoc extradoc: Slides
arigo
- [pypy-commit] extradoc extradoc: tweaks
arigo
- [pypy-commit] extradoc extradoc: updated slides, check in the html version
arigo
- [pypy-commit] extradoc extradoc: Update people.txt
arigo
- [pypy-commit] extradoc extradoc: updates
arigo
- [pypy-commit] extradoc extradoc: updates
arigo
- [pypy-commit] extradoc extradoc: ups, did not read the line above, removed me from the list of prev. attendees
plan_rich
- [pypy-commit] pypy.org extradoc: Add a warning: pypy3 is slow and old
arigo
- [pypy-commit] pypy.org extradoc: Mention the "@pypy-4.0.1" part of the url
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy cffi-embedding-win32: first stab at fixing embedding for win32
mattip
- [pypy-commit] pypy cffi-static-callback-embedding: Add missing files
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: fix
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: Fixes. Yay! The cffi tests (after translation) pass.
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: fix fix fix
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: Forgot to "hg add" these new tests
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge default
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge default (including the ec-keepalive branch)
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge ec-keepalive
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge ec-keepalive
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge ec-keepalive
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: hg merge ec-keepalive
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: merge heads
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: ready to merge
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: redirect readers from embedding.rst to cffi's own embedding.rst
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: update to cffi's current static-callback-embedding branch
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: update to cffi/1c789b36273d
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: update to cffi/4ce79f707838
arigo
- [pypy-commit] pypy cffi-static-callback-embedding: update to cffi/abd64e2e97cb
arigo
- [pypy-commit] pypy cpyext-bootstrap: A branch to simplify the bootstrapping, to reduce the amount of changes
arigo
- [pypy-commit] pypy cpyext-bootstrap: Attempt to fix translation, getting segfault while translating now
arigo
- [pypy-commit] pypy cpyext-bootstrap: Kill kill kill the logic in init_typeobject(). It was anyway very
arigo
- [pypy-commit] pypy cpyext-bootstrap: pfff, took me a while to figure out that memory corruption was going on here:
arigo
- [pypy-commit] pypy cpyext-bootstrap: Seems to work, merging
arigo
- [pypy-commit] pypy cpyext-bootstrap: Some translation fixes
arigo
- [pypy-commit] pypy cpyext-bootstrap: This test checks a condition which is actually not true, but which
arigo
- [pypy-commit] pypy cpyext-ext: add a make_typedescr for W_MemeberDesc
mattip
- [pypy-commit] pypy cpyext-ext: add a TypeDef to W_MemberDesc with the corrent name
mattip
- [pypy-commit] pypy cpyext-ext: add make_typedescr for W_GetSetPropertyEx <-> PyGetSetDescrObject conversion
mattip
- [pypy-commit] pypy cpyext-ext: add passing test for multiple inheritance
mattip
- [pypy-commit] pypy cpyext-ext: add protective hack to work around PyString_Type missing a ob_type reference, give up writing a test to properly fix it
mattip
- [pypy-commit] pypy cpyext-ext: avoid segfault by detecting the problem in c, move the python-level test to fail first
mattip
- [pypy-commit] pypy cpyext-ext: copy add_docstring from numpy, add test that passes with -A
mattip
- [pypy-commit] pypy cpyext-ext: export only what needs to be exported
mattip
- [pypy-commit] pypy cpyext-ext: fix ref leak in test
mattip
- [pypy-commit] pypy cpyext-ext: fix translation
mattip
- [pypy-commit] pypy cpyext-ext: inherit tp_as... slots from base, need to test?
mattip
- [pypy-commit] pypy cpyext-ext: make test more complete, now fails
mattip
- [pypy-commit] pypy cpyext-ext: make the header compatible with cpython
mattip
- [pypy-commit] pypy cpyext-ext: merge default into branch
mattip
- [pypy-commit] pypy cpyext-ext: merge default into branch
mattip
- [pypy-commit] pypy cpyext-ext: reduce optimization level in test module compilation
mattip
- [pypy-commit] pypy cpyext-ext: remove debug cruft
mattip
- [pypy-commit] pypy cpyext-ext: rewrite the test, do not modify a readonly attribute; stcmp() instead. Exempt checking for getset, type objects
mattip
- [pypy-commit] pypy cpyext-ext: test, fix - raise cpython compatible exception
mattip
- [pypy-commit] pypy cpyext-gc-support-2: Add a comment (thanks ronan)
arigo
- [pypy-commit] pypy cpyext-gc-support-2: Copy more code from cpyext-gc-support
arigo
- [pypy-commit] pypy cpyext-gc-support-2: copy some code, rekill parts that I definitely want killed
arigo
- [pypy-commit] pypy cpyext-gc-support-2: copy the basic GC support
arigo
- [pypy-commit] pypy cpyext-gc-support-2: fix
arigo
- [pypy-commit] pypy cpyext-gc-support-2: fixes, now getting a segfault
arigo
- [pypy-commit] pypy cpyext-gc-support-2: fix imports
arigo
- [pypy-commit] pypy cpyext-gc-support-2: Found the segfault's origin, made more explicit that way (not fixed yet)
arigo
- [pypy-commit] pypy cpyext-gc-support-2: hg merge default
arigo
- [pypy-commit] pypy cpyext-gc-support-2: Hopefully fix the bootstrap cycles in this way
arigo
- [pypy-commit] pypy cpyext-gc-support-2: in-progress
arigo
- [pypy-commit] pypy cpyext-gc-support-2: in-progress
arigo
- [pypy-commit] pypy cpyext-gc-support-2: more copying
arigo
- [pypy-commit] pypy cpyext-gc-support-2: More tweaks for bootstrapping
arigo
- [pypy-commit] pypy cpyext-gc-support-2: probably fixed by moving this part of the code later (this is part of an
arigo
- [pypy-commit] pypy cpyext-gc-support-2: Rename: this new function should really be a superset of make_ref()
arigo
- [pypy-commit] pypy cpyext-gc-support-2: rename from_pyobj() back to from_ref() too
arigo
- [pypy-commit] pypy cpyext-gc-support-2: Restarting the cpyext-gc-support branch from scratch
arigo
- [pypy-commit] pypy cpyext-gc-support-2: updates
arigo
- [pypy-commit] pypy cpyext-gc-support-2: yay, test_api passes
arigo
- [pypy-commit] pypy default: 32-bit fix
arigo
- [pypy-commit] pypy default: Actually remove all references to GETFIELD_GC_PURE_* opcode
sbauman
- [pypy-commit] pypy default: Actually remove GETFIELD_GC_PURE_* from the opcode space
sbauman
- [pypy-commit] pypy default: Add "-m ensurepip"
arigo
- [pypy-commit] pypy default: Add __pypy__.decode_long(), an app-level interface to
arigo
- [pypy-commit] pypy default: Add slots here, to make sure that nobody is using "self.storage" any longer
arigo
- [pypy-commit] pypy default: Add support for floats as parameters to itertools.islice. Improves compatibility with CPython.
jonaspf
- [pypy-commit] pypy default: another assert: we only re-size the ._items of an array info if it is not
cfbolz
- [pypy-commit] pypy default: a whatsnew entry
cfbolz
- [pypy-commit] pypy default: Be clearer about that
arigo
- [pypy-commit] pypy default: Better use %r than '%s' in case it's a non-printable byte
arigo
- [pypy-commit] pypy default: Cast to PyObject* in PyString_GET_SIZE/AS_STRING.
devin.jeanpierre
- [pypy-commit] pypy default: cffi_imports depends on pypy-c being built first
arigo
- [pypy-commit] pypy default: Change GraphAnalyzer to use a more precise way to recognize external functions.
mjacob
- [pypy-commit] pypy default: Change this part of the test to use an actual external function.
mjacob
- [pypy-commit] pypy default: Clean up after 58ef780a3875.
mjacob
- [pypy-commit] pypy default: Cleanup get_graph() function.
mjacob
- [pypy-commit] pypy default: cpyext implies micronumpy, otherwise we get never-initialized ndarray subclasses
arigo
- [pypy-commit] pypy default: CPython has a special case for ``long("string", power-of-two-base)`` to
arigo
- [pypy-commit] pypy default: Crash with the same UnpicklingError as CPython when asked to unpickle
arigo
- [pypy-commit] pypy default: Document branch 'exctrans'
rlamy
- [pypy-commit] pypy default: Document the branch
arigo
- [pypy-commit] pypy default: don't create 2-tuples all the time in mapdict. pass the content along instead
cfbolz
- [pypy-commit] pypy default: duh, check the index first(!)
cfbolz
- [pypy-commit] pypy default: Expand the error message
arigo
- [pypy-commit] pypy default: Expose SOABI in sysconfig
stefanor
- [pypy-commit] pypy default: fix
arigo
- [pypy-commit] pypy default: fix - link_flags is a tuple not a list
mattip
- [pypy-commit] pypy default: Fix: we need to "acquire/release the GIL" even around RPython_Startup(), like entrypoint.c does
arigo
- [pypy-commit] pypy default: Fix _PyLong_FromByteArray() for e205bcf52d2f
arigo
- [pypy-commit] pypy default: Fix bogus assembler generated in a case that never occurs in practice
arigo
- [pypy-commit] pypy default: fix link, add more
arigo
- [pypy-commit] pypy default: Fix test_rabspath_absolute_nt
Vincent Legoll
- [pypy-commit] pypy default: Fix the complexity in cPickle.py too (there's a mostly-duplicate?)
arigo
- [pypy-commit] pypy default: Fix the test. _PyLong_FromByteArray() always produces nonsense (i.e. a
arigo
- [pypy-commit] pypy default: Fix two tests in kwargsdict that have a typo
arigo
- [pypy-commit] pypy default: fix typo
cfbolz
- [pypy-commit] pypy default: have consistent argument order between setfield and setitem
cfbolz
- [pypy-commit] pypy default: hg merge cffi-static-callback-embedding
arigo
- [pypy-commit] pypy default: hg merge cpyext-bootstrap
arigo
- [pypy-commit] pypy default: hg merge ec-keepalive
arigo
- [pypy-commit] pypy default: ignore branch
arigo
- [pypy-commit] pypy default: Issue #2222: test and fix
arigo
- [pypy-commit] pypy default: kill unused methods
cfbolz
- [pypy-commit] pypy default: Let GraphAnalyzer's analyze_external_call() method take a function object instead of the call operation.
mjacob
- [pypy-commit] pypy default: Let GraphAnalyzer's analyze_external_call() return a more conservative result in case of a delayed pointer.
mjacob
- [pypy-commit] pypy default: Let GraphAnalyzer return a conservative result if it encounters a function object which is neither external nor has a graph attached.
mjacob
- [pypy-commit] pypy default: Let GraphAnalyzer return a conservative result instead of crashing if it encounters a call of a delayed pointer.
mjacob
- [pypy-commit] pypy default: Link to http://pypy.org/download.html#building-from-source, which
arigo
- [pypy-commit] pypy default: Make C-API char* parameters const-correct with respect to CPython.
devin.jeanpierre
- [pypy-commit] pypy default: Make GraphAnalyzer return the bottom result when encountering a null function pointer.
mjacob
- [pypy-commit] pypy default: Meh! The rbigint class has already got a frombytes() method, which
arigo
- [pypy-commit] pypy default: Mention here the _PyLong_FromByteArray() fix
arigo
- [pypy-commit] pypy default: Mention remove-getfield-pure branch in whatsnew-head
sbauman
- [pypy-commit] pypy default: merge
cfbolz
- [pypy-commit] pypy default: merge
cfbolz
- [pypy-commit] pypy default: Merged in exctrans (pull request #390)
rlamy
- [pypy-commit] pypy default: Merged in vincentlegoll/pypy/fix-cpython-ssl-tests-2.7 (pull request #389)
fijal
- [pypy-commit] pypy default: merge globals-quasiimmut:
cfbolz
- [pypy-commit] pypy default: merge heads
arigo
- [pypy-commit] pypy default: Merge latest changes
sbauman
- [pypy-commit] pypy default: Merge remove-getfield-pure
sbauman
- [pypy-commit] pypy default: Minor performance improvement
arigo
- [pypy-commit] pypy default: Missing headers
arigo
- [pypy-commit] pypy default: More tests, more fixes
arigo
- [pypy-commit] pypy default: move file
arigo
- [pypy-commit] pypy default: Note
arigo
- [pypy-commit] pypy default: one missed "strategy" field read
arigo
- [pypy-commit] pypy default: put the really shared method into an abstract base class
cfbolz
- [pypy-commit] pypy default: reduce the confusion of everything being name "descr" by renaming a few "descr"
cfbolz
- [pypy-commit] pypy default: remove all traces of micronumpy from cpyext if --withoutmod-micronumpy option used
mattip
- [pypy-commit] pypy default: Remove handling of DelayedPointer in all analyze_external_call() methods.
mjacob
- [pypy-commit] pypy default: remove some unused remnants from before the big refactoring
cfbolz
- [pypy-commit] pypy default: Remove the need for passing "config" to StackletThread().
arigo
- [pypy-commit] pypy default: rename _setfield to put_field_back_to_info and document it (it's only called
cfbolz
- [pypy-commit] pypy default: rename lazy_setfield to lazy_set (since it also means "setarrayitem")
cfbolz
- [pypy-commit] pypy default: rename register_dirty_field to register_info. document and check its invariant.
cfbolz
- [pypy-commit] pypy default: rename the really generic _getvalue to _get_rhs_from_set_op
cfbolz
- [pypy-commit] pypy default: review the skips in test_optimizebasic
cfbolz
- [pypy-commit] pypy default: Simplify a bit the code here
arigo
- [pypy-commit] pypy default: Simplify code.
mjacob
- [pypy-commit] pypy default: skip for msvc
mattip
- [pypy-commit] pypy default: slightly better test
arigo
- [pypy-commit] pypy default: Test and fix
arigo
- [pypy-commit] pypy default: test and fix for the sandboxing issue on 32-bit (test_lseek)
arigo
- [pypy-commit] pypy default: this comment is outdated now
cfbolz
- [pypy-commit] pypy default: those just pass
cfbolz
- [pypy-commit] pypy default: typo
arigo
- [pypy-commit] pypy default: Typo (thanks Vincent)
arigo
- [pypy-commit] pypy default: Update the date
arigo
- [pypy-commit] pypy default: update whatsnew-head.rst
arigo
- [pypy-commit] pypy ec-keepalive: - fix the 'ready = 0', which should be after memset(), otherwise it
arigo
- [pypy-commit] pypy ec-keepalive: Add custom trace hooks in order to walk all threadlocalrefs,
arigo
- [pypy-commit] pypy ec-keepalive: Add objectmodel.fetch_translated_config()
arigo
- [pypy-commit] pypy ec-keepalive: After fork(), at least in the __thread case, we need to be careful and
arigo
- [pypy-commit] pypy ec-keepalive: Can't use this way of declaring structures with MSVC
arigo
- [pypy-commit] pypy ec-keepalive: Document and cross-reference these two modules, which implement
arigo
- [pypy-commit] pypy ec-keepalive: Explicitly disable automatic_keepalive() in one case
arigo
- [pypy-commit] pypy ec-keepalive: Finish the RPython part
arigo
- [pypy-commit] pypy ec-keepalive: fix
arigo
- [pypy-commit] pypy ec-keepalive: fixes
arigo
- [pypy-commit] pypy ec-keepalive: Force a 'shared' build in these tests, for Windows
arigo
- [pypy-commit] pypy ec-keepalive: hg merge default
arigo
- [pypy-commit] pypy ec-keepalive: Maybe it's a better idea to use critical sections here
arigo
- [pypy-commit] pypy ec-keepalive: merge heads
arigo
- [pypy-commit] pypy ec-keepalive: Missing includes
arigo
- [pypy-commit] pypy ec-keepalive: Potential fix: need to use __sync_lock_release() here
arigo
- [pypy-commit] pypy ec-keepalive: Progress, see comments
arigo
- [pypy-commit] pypy ec-keepalive: ready for merge
arigo
- [pypy-commit] pypy ec-keepalive: Simplify: rshrinklist keeps the order, unlike rweaklist which keeps the
arigo
- [pypy-commit] pypy ec-keepalive: Trying to keep alive the ExecutionContext built in new threads for
arigo
- [pypy-commit] pypy ec-keepalive: Yet Another attempt to fix rpython_startup_code(), this time writing it
arigo
- [pypy-commit] pypy exctrans: Backed out changeset 2906290a8ebc
rlamy
- [pypy-commit] pypy exctrans: Call db.prepare_inline_helpers() from a slightly more logical location
rlamy
- [pypy-commit] pypy exctrans: Call funcgen.patch_graph() before source generation rather than in the middle of it
rlamy
- [pypy-commit] pypy exctrans: cleanup
rlamy
- [pypy-commit] pypy exctrans: Close branch exctrans
rlamy
- [pypy-commit] pypy exctrans: Create factory function for creating FuncNodes
rlamy
- [pypy-commit] pypy exctrans: Create sandbox stubs in the annotator and finally get rid of databasing-time creation of sandbox helpers
rlamy
- [pypy-commit] pypy exctrans: Don't create the database automagically in .generate_graphs()
rlamy
- [pypy-commit] pypy exctrans: Do sandboxing of os.* replacements at the rtyper level.
rlamy
- [pypy-commit] pypy exctrans: Do sandboxing of register_external() stuff at the rtyper level
rlamy
- [pypy-commit] pypy exctrans: Do the sandbox transform by hacking fnobj.__dict__
rlamy
- [pypy-commit] pypy exctrans: Do the sandbox transform slightly earlier
rlamy
- [pypy-commit] pypy exctrans: Extract function
rlamy
- [pypy-commit] pypy exctrans: Extract name computation out of FuncNode.__init__
rlamy
- [pypy-commit] pypy exctrans: extract sandboxing case in select_function_code_generators()
rlamy
- [pypy-commit] pypy exctrans: fix database tests
rlamy
- [pypy-commit] pypy exctrans: fix tests
rlamy
- [pypy-commit] pypy exctrans: hg merge default
rlamy
- [pypy-commit] pypy exctrans: hg merge default
rlamy
- [pypy-commit] pypy exctrans: inline make_funcgens()
rlamy
- [pypy-commit] pypy exctrans: kill patch_graph()
rlamy
- [pypy-commit] pypy exctrans: Kill trivial method funcgen.name()
rlamy
- [pypy-commit] pypy exctrans: merge generate_graphs() into db.complete()
rlamy
- [pypy-commit] pypy exctrans: merge prepare_inline_helpers() and inline_helpers(); simplify
rlamy
- [pypy-commit] pypy exctrans: move db.need_sandboxing to the only module using it
rlamy
- [pypy-commit] pypy exctrans: Remove dead code
rlamy
- [pypy-commit] pypy exctrans: Simplify: split off ExternalFuncNode class from FuncNode
rlamy
- [pypy-commit] pypy exctrans: simplify code
rlamy
- [pypy-commit] pypy exctrans: simplify code: always patch the graph in-place in funcgen
rlamy
- [pypy-commit] pypy exctrans: Simplify get_external_function_sandbox_graph() a bit
rlamy
- [pypy-commit] pypy exctrans: Split get_external_function_sandbox_graph() and kill force_stub flag
rlamy
- [pypy-commit] pypy fix-3.3-pypy-magic: When doing init_builtin, force a module.init() call to make sure we don't just grab the module out of sys.modules. (Fixes reload())
marky1991
- [pypy-commit] pypy fix-cpython-ssl-tests-2.7: Add SSL fixing branch description
Vincent Legoll
- [pypy-commit] pypy fix-cpython-ssl-tests-2.7: Fix cpython test suite's SSL tests
Vincent Legoll
- [pypy-commit] pypy fix-cpython-ssl-tests-2.7: New branch to fix SSL tests by porting the cpython patch
Vincent Legoll
- [pypy-commit] pypy globals-quasiimmut: close to-be-merged branch
cfbolz
- [pypy-commit] pypy globals-quasiimmut: frame.f_globals is not writable
cfbolz
- [pypy-commit] pypy globals-quasiimmut: introduce a special class W_ModuleDictObject which stores the strategy as a
cfbolz
- [pypy-commit] pypy globals-quasiimmut: rejigger test_pypy_c tests
cfbolz
- [pypy-commit] pypy globals-quasiimmut: remove confusing class level default
cfbolz
- [pypy-commit] pypy globals-quasiimmut: store the first w_globals that a code object is run in on the code object. if a
cfbolz
- [pypy-commit] pypy jit-leaner-frontend: a bit of random progress
fijal
- [pypy-commit] pypy llvm-translation-backend: Add a hack to make functions linked in from C code inlinable.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Explicitly set random_effects_on_gcobjs flag if constructing an external function pointer.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Fix this test by using the new entrypoint API.
mjacob
- [pypy-commit] pypy llvm-translation-backend: hg merge default
mjacob
- [pypy-commit] pypy llvm-translation-backend: hg merge default
mjacob
- [pypy-commit] pypy llvm-translation-backend: hg merge default
mjacob
- [pypy-commit] pypy llvm-translation-backend: Implement a LLVM plugin pass which internalizes hidden symbols.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Implement gc_thread_run() operation.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Move code that prepares graph before writing in its own method.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Put gc_reload_possibly_moved transformation in its own method.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Remove dead code.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Remove parts of the llvmgcroot implementation.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Set visibility of call wrappers to 'hidden'.
mjacob
- [pypy-commit] pypy llvm-translation-backend: Shuffle around some code in anticipation of the new llvmgcroot implementation.
mjacob
- [pypy-commit] pypy memop-simplify3: added what is new entry
plan_rich
- [pypy-commit] pypy memop-simplify3: merged default
plan_rich
- [pypy-commit] pypy memop-simplify3: scaling the length loc in rewrite for malloc_cond_varsize
plan_rich
- [pypy-commit] pypy memop-simplify3: syntax error removed, simplified op.getopnum() to opnum and extracting the field before
plan_rich
- [pypy-commit] pypy memop-simplify3: translation fixes for the changes in rewrite.py
plan_rich
- [pypy-commit] pypy osx-vmprof-support: close branch likely not going anywhere (different approach ahead)
fijal
- [pypy-commit] pypy py3.3-marky1991-fix-ssl: Fixed ssl.
marky1991
- [pypy-commit] pypy py3.3: Add _ssl.RAND_bytes
amauryfa
- [pypy-commit] pypy py3.3: Add pickle support to itertools.combinations
amauryfa
- [pypy-commit] pypy py3.3: Add pickle support to itertools.cycle
amauryfa
- [pypy-commit] pypy py3.3: Add test to see if sys.exceptionhook is used when outputting errors in code.runsource
marky1991
- [pypy-commit] pypy py3.3: All the pickle-related tests still seem to be passing after making less drastic changes to pickle.py .
marky1991
- [pypy-commit] pypy py3.3: Another test that needed a change for CPython3.3
amauryfa
- [pypy-commit] pypy py3.3: As of 3.3, this scenario actually raises a KeyError on Cpython, not an ImportError anymore.
marky1991
- [pypy-commit] pypy py3.3: Cleanup handling of __loader__ in test_mod
marky1991
- [pypy-commit] pypy py3.3: Cleanup newline in test_import_lock to match base stdlib.
marky1991
- [pypy-commit] pypy py3.3: Close this head.
mjacob
- [pypy-commit] pypy py3.3: Fix class name.
amauryfa
- [pypy-commit] pypy py3.3: Fix lib_pypy/test_code_module. Still need to make sure that the tests here are tested elsewhere.
marky1991
- [pypy-commit] pypy py3.3: Fix many many pickle-related failures. The changes to frozen_importlib are questionable and might be wrong. (Once you replace __import__ with import_with_frames_removed, when we try to pickle __import__ it fails to find a function with name __import__, causing pickling to fail.) Because unbound methods bacame regular functions in python3, pickling them triggers save_global, which fails because they're obviously not globals. Thus, I changed pickle.py to just use the function's __reduce/reduce_ex__ methods as they do in the default branch.
marky1991
- [pypy-commit] pypy py3.3: Fix pickling stuff. Also, when (un)pickling functions, pass qualname correctly.
marky1991
- [pypy-commit] pypy py3.3: Fix rposix module after bad merge
amauryfa
- [pypy-commit] pypy py3.3: Fix sys module testing.
marky1991
- [pypy-commit] pypy py3.3: Fix tests in stderrprinter and io.
marky1991
- [pypy-commit] pypy py3.3: Fix translation.
mjacob
- [pypy-commit] pypy py3.3: Fix translation error caused by the qualname fix.
marky1991
- [pypy-commit] pypy py3.3: Fix unicode.capitalize() test to pass with CPython3.3,
amauryfa
- [pypy-commit] pypy py3.3: For some reason this chunk was not copied from CPython 3.3
amauryfa
- [pypy-commit] pypy py3.3: Get rid of debugging.
marky1991
- [pypy-commit] pypy py3.3: Get rid of debugging crap from the tests for import locks.
marky1991
- [pypy-commit] pypy py3.3: Get rid of debugging junk.
marky1991
- [pypy-commit] pypy py3.3: hg merge 03591a1499c8
mjacob
- [pypy-commit] pypy py3.3: hg merge py3k (+ fixes)
mjacob
- [pypy-commit] pypy py3.3: Importlib raises a TypeError when you try to set __package__ to a string nowadays. Updating the test.
marky1991
- [pypy-commit] pypy py3.3: io.BufferedWriter: completely remove max_buffer_size argument.
amauryfa
- [pypy-commit] pypy py3.3: Make sure to grab the import lock when importing. Not sure if this a sufficent fix, but it at least adds this back.
marky1991
- [pypy-commit] pypy py3.3: Make the diff from the stdlib smaller.
marky1991
- [pypy-commit] pypy py3.3: Match commenting style to PEP-8 as requested. Also, went ahead and fixed the code module to mostly match py3k.
marky1991
- [pypy-commit] pypy py3.3: Merge ssl fixes back into py3.3
marky1991
- [pypy-commit] pypy py3.3: Merging magic fixes into py3.3 .
marky1991
- [pypy-commit] pypy py3.3: Merging master py3.3 into my fork.
marky1991
- [pypy-commit] pypy py3.3: Pickle support for itertools.islice
amauryfa
- [pypy-commit] pypy py3.3: pickle support for itertools.takewhile and dropwhile.
amauryfa
- [pypy-commit] pypy py3.3: Redo 88aafcb7c318: hg merge py3k (+ fixes)
mjacob
- [pypy-commit] pypy py3.3: Relax a test: with pypy we cannot choose the name of the function when argument parsing fails.
amauryfa
- [pypy-commit] pypy py3.3: Remove final unnecessary change to test_import_lock
marky1991
- [pypy-commit] pypy py3.3: Remove this test as per 4d306d6.
marky1991
- [pypy-commit] pypy py3.3: Remove trailing whitespace.
mjacob
- [pypy-commit] pypy py3.3: Remove unnecessary comments.
marky1991
- [pypy-commit] pypy py3.3: Remove unused frozen_importlib import.
marky1991
- [pypy-commit] pypy py3.3: Revert change to cffi backend test.
marky1991
- [pypy-commit] pypy py3.3: Revert unnecessary change to frozen importlib. The changes to the general pickle module make this no longer needed.
marky1991
- [pypy-commit] pypy py3.3: TextIOWrapper: Ensure the constructor complains if passed a codec that isn't
amauryfa
- [pypy-commit] pypy py3.3: Turn qualname into a unicode object in the ast phase. (Fixes repr/str of class objects with non-ascii names)
marky1991
- [pypy-commit] pypy py3.3: Turn set(char) into set([char]) as per review.
marky1991
- [pypy-commit] pypy py3.3: unicodedata: add tables for special_casing.
amauryfa
- [pypy-commit] pypy py3.3: Update cffi backend test c file.
marky1991
- [pypy-commit] pypy py3k: Fix import for renamed module rposix -> rposix_stat.
mjacob
- [pypy-commit] pypy remove-getfield-pure: Check no longer needed
sbauman
- [pypy-commit] pypy remove-getfield-pure: Cleanup based on suggestions from fijal
sbauman
- [pypy-commit] pypy remove-getfield-pure: Cleanup logic and include GETFIELDS in list of descriptor based pure operations
sbauman
- [pypy-commit] pypy remove-getfield-pure: Directly consult descr for getfield operations
sbauman
- [pypy-commit] pypy remove-getfield-pure: Don't invalidate field caches for immutable fields
sbauman
- [pypy-commit] pypy remove-getfield-pure: Forgot to change this back earlier
sbauman
- [pypy-commit] pypy remove-getfield-pure: Log getfield operations as pure baesd on their descriptor
sbauman
- [pypy-commit] pypy remove-getfield-pure: Merge with default
sbauman
- [pypy-commit] pypy remove-getfield-pure: Merge with default
sbauman
- [pypy-commit] pypy remove-getfield-pure: Merge with default
sbauman
- [pypy-commit] pypy remove-getfield-pure: Merge with default
sbauman
- [pypy-commit] pypy remove-getfield-pure: More precise logic for constant folding
sbauman
- [pypy-commit] pypy remove-getfield-pure: Put purity information of field into repr of its descriptor
sbauman
- [pypy-commit] pypy remove-getfield-pure: Remove '_pure' suffix from getfield operations (moving it to descr)
sbauman
- [pypy-commit] pypy remove-getfield-pure: Remove getfield_gc_pure_* instructions from tests (purity information is now in the descriptor)
sbauman
- [pypy-commit] pypy remove-getfield-pure: Test case for a large number of immutable field references
sbauman
- [pypy-commit] pypy remove-getfield-pure: Treat quasiimmut fields as pure for optimization purposes
sbauman
- [pypy-commit] pypy repeatlist_strategy: Add support to skip tests if some list strategies are available in the interpreter
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: backout e2d983c38082, it was wrong, the failure comes from repeatlist strategy
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: Create a new branch to test the repeatlist strategy
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: Fix test, was:
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: Initial commit
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: merge default
Vincent Legoll
- [pypy-commit] pypy repeatlist_strategy: merge default
Vincent Legoll
- [pypy-commit] pypy s390x-backend: adapted test to match the expected assembly after the changes
plan_rich
- [pypy-commit] pypy s390x-backend: adapted values unpacked from the gcmap for the s390x case
plan_rich
- [pypy-commit] pypy s390x-backend: added assertion to pool, rewrote assembler of int_mul_ovf
plan_rich
- [pypy-commit] pypy s390x-backend: added case where parameters are not ordered
plan_rich
- [pypy-commit] pypy s390x-backend: added comment to my last commit
plan_rich
- [pypy-commit] pypy s390x-backend: added missing parameter to LMG call (refactoring issues :)
plan_rich
- [pypy-commit] pypy s390x-backend: added more failing tests for the pool, filling the literal pool after the trace list has been rewritten (did not switch the order for bridges, fixes more tests)
plan_rich
- [pypy-commit] pypy s390x-backend: added nursery_ptr_increment, added lldebug to some tests to debug in gdb
plan_rich
- [pypy-commit] pypy s390x-backend: added overflow test of pool (not yet finished)
plan_rich
- [pypy-commit] pypy s390x-backend: added SGRK instruction, now the stack overflow is correctly detected and the exception is thrown!
plan_rich
- [pypy-commit] pypy s390x-backend: added some sanity checks, s390x only fails 4 (in test_runner)
plan_rich
- [pypy-commit] pypy s390x-backend: added test_basic & test_calling_conventions, the latter already passes
plan_rich
- [pypy-commit] pypy s390x-backend: added test_float from backend/test (passing)
plan_rich
- [pypy-commit] pypy s390x-backend: added test recursive, four tests to fix
plan_rich
- [pypy-commit] pypy s390x-backend: added tests: loop_unroll, virtualizable, virtualref
plan_rich
- [pypy-commit] pypy s390x-backend: added to more tests & bug fix: do not use SRLG (logical right shift) or any other shift with parameters R_1 and R_3 that designate the same register
plan_rich
- [pypy-commit] pypy s390x-backend: added zrpy_gc test file,
plan_rich
- [pypy-commit] pypy s390x-backend: added ztranslation call assembler test and some more (hurray, one of them fails and this might be the bug I'm searching for)
plan_rich
- [pypy-commit] pypy s390x-backend: adding -march=zEC12 to the platform's cflags (linux)
plan_rich
- [pypy-commit] pypy s390x-backend: advancing the pointer to correctly return the value written from libffi, the reason why this does not happen on e.g. ppc bigendian 64 bit is: libffi ppc casts the return value to the requested type (s390x does not and we have discussed this enough already)
plan_rich
- [pypy-commit] pypy s390x-backend: and the other part of the bigendian issues (micronumpy tests)
plan_rich
- [pypy-commit] pypy s390x-backend: and yet some more tests: del, dict, exception and fficall
plan_rich
- [pypy-commit] pypy s390x-backend: another case where regalloc pair did not succeed
plan_rich
- [pypy-commit] pypy s390x-backend: asmlen test failed, because more instructions in the entry of a bridge are now compiled (realloc check), seems about right what other backends implement!
plan_rich
- [pypy-commit] pypy s390x-backend: big endian issue while testing cpyext (PyUnicode_DecodeUTF16)
plan_rich
- [pypy-commit] pypy s390x-backend: big endian test issue cpyext
plan_rich
- [pypy-commit] pypy s390x-backend: dict comprehension rewrite to loop overwrote parameter variable of the function
plan_rich
- [pypy-commit] pypy s390x-backend: did not think of the following scenario: in a bridge you cannot 'just' use volatile registers as some might be allocated in the trace it is exiting (fixed)
plan_rich
- [pypy-commit] pypy s390x-backend: did not write pool entires to the right location
plan_rich
- [pypy-commit] pypy s390x-backend: do not assert, but check if it is a CallDescr, in the case call_assembler, assume the default word size
plan_rich
- [pypy-commit] pypy s390x-backend: do not use a register that might be allocated in copy_content! added comment to clarify. test_string is now passing
plan_rich
- [pypy-commit] pypy s390x-backend: due to the refacotring the variable saving went to the wrong stack frame in wb_slowpath (now takes the right one)
plan_rich
- [pypy-commit] pypy s390x-backend: endian test issue marshal, test decoded value using platform endianess (not desired for marshall module)
plan_rich
- [pypy-commit] pypy s390x-backend: ensure the index arg of gc_load_indexed to be in a register (assembler assumed this), adding offset to _rewrite_changeable_constptrs
plan_rich
- [pypy-commit] pypy s390x-backend: exchanged AGHI with LAY, same effect, but LAY does not change the condition code which is needed by the write barrier helper!
plan_rich
- [pypy-commit] pypy s390x-backend: excluding structs from buffer advance (which would be wrong)
plan_rich
- [pypy-commit] pypy s390x-backend: extended the stack limit of s390x
plan_rich
- [pypy-commit] pypy s390x-backend: fixed 2 translation issues
plan_rich
- [pypy-commit] pypy s390x-backend: fixed another 3 tests (translated call release gil)
plan_rich
- [pypy-commit] pypy s390x-backend: fixed a pool issue where a guard token did not receive it's rightful pool position
plan_rich
- [pypy-commit] pypy s390x-backend: fixed callsite of clibffi with the same big endian issues as found yesterday evening
plan_rich
- [pypy-commit] pypy s390x-backend: fixed problem in build_propagate_exception_path, value was loaded into wrong register (that was overwritten after that immediatley)
plan_rich
- [pypy-commit] pypy s390x-backend: fixed syntax error
plan_rich
- [pypy-commit] pypy s390x-backend: fixed tests that fail after applying the scale in gc_load_indexed_r (llsupport/gc.py)
plan_rich
- [pypy-commit] pypy s390x-backend: fixed translation issue. reverted the stacklet changes (did not pass test)
plan_rich
- [pypy-commit] pypy s390x-backend: gotcha, guard nonnull class was implemented incorrectly (substracted value that should have been shifted 16 bits to the left)
plan_rich
- [pypy-commit] pypy s390x-backend: greater equal is not equal, this basically invoked realloc frame EVERY time the assembler location was visited
plan_rich
- [pypy-commit] pypy s390x-backend: guard_subclass, guard_is_object ported to s390x
plan_rich
- [pypy-commit] pypy s390x-backend: half way through the ndarray tests (endian issues)
plan_rich
- [pypy-commit] pypy s390x-backend: implemented stacklet switch (slp_switch) for s390x
plan_rich
- [pypy-commit] pypy s390x-backend: int_lshift is a logical one, but up to now emitted an arithmetic shift, this makes the test_basic of the s390x fully passing!
plan_rich
- [pypy-commit] pypy s390x-backend: issue while rewriting, missed constant factor that is not multiplied to index
plan_rich
- [pypy-commit] pypy s390x-backend: it can happen that longevity does not contain an entry for an operation (e.g. int_mul_ovf and result is not used), then when trying to spill a variable op can be in reg_bindings, but is not in longevity -> KeyError,
plan_rich
- [pypy-commit] pypy s390x-backend: jumping over pair (end of regalloc) if one of them is forbidden
plan_rich
- [pypy-commit] pypy s390x-backend: libffi issue handled just after calling. narrower integer types trigger zero/sign extension
plan_rich
- [pypy-commit] pypy s390x-backend: literal pool enhancement. it now stores unique values, no 64 bit integer/float/ref is every stored twice in the pool
plan_rich
- [pypy-commit] pypy s390x-backend: load_imm might emit different code load 32 or 64 bit imm, added this case to the regex of test_compile_asmlen
plan_rich
- [pypy-commit] pypy s390x-backend: macros (e.g. WCOREDUMP) got parameter type Signed, on little endian this does not make a difference, but it does on big endian. changed to rffi.INT
plan_rich
- [pypy-commit] pypy s390x-backend: made register r13 dirty before flushing the non volatile registers to the stack,
plan_rich
- [pypy-commit] pypy s390x-backend: malloc_cond_varsize_frame ported to s390x
plan_rich
- [pypy-commit] pypy s390x-backend: malloc_cond_varsize impl
plan_rich
- [pypy-commit] pypy s390x-backend: malloc_cond ported to s390x
plan_rich
- [pypy-commit] pypy s390x-backend: malloc_nursery_* fixes, need to revive the shadowstack calls (e.g. call_footer_shadow_stack) to pass the full gc integration test
plan_rich
- [pypy-commit] pypy s390x-backend: merged default
plan_rich
- [pypy-commit] pypy s390x-backend: merged default
plan_rich
- [pypy-commit] pypy s390x-backend: merged default
plan_rich
- [pypy-commit] pypy s390x-backend: merged default
plan_rich
- [pypy-commit] pypy s390x-backend: merged default
plan_rich
- [pypy-commit] pypy s390x-backend: merged head
plan_rich
- [pypy-commit] pypy s390x-backend: merged memop simplify (malloc_nursery_varsize updates)
plan_rich
- [pypy-commit] pypy s390x-backend: merged memop simplify, added whatsnew entry to this branch
plan_rich
- [pypy-commit] pypy s390x-backend: missing files
plan_rich
- [pypy-commit] pypy s390x-backend: missing import for llop
plan_rich
- [pypy-commit] pypy s390x-backend: modified 2 tests, first used void pointer instead of int pointer (thus qsort did not sort),
plan_rich
- [pypy-commit] pypy s390x-backend: more than one pool constant did not correctly increment the pool cursor
plan_rich
- [pypy-commit] pypy s390x-backend: mul overflow, if right is negative up to now the wrong branch of mul overflow has been taken. resulted in abs(v) = abs(x) * neg(z), which is wrong
plan_rich
- [pypy-commit] pypy s390x-backend: need to save all registers before assembling call_release_gil*, accidentally put false to save_all_regs
plan_rich
- [pypy-commit] pypy s390x-backend: not every call needs to reload the pool register, but call_assembler must perform a reload
plan_rich
- [pypy-commit] pypy s390x-backend: numpy dtype fixes in the test suite
plan_rich
- [pypy-commit] pypy s390x-backend: pair regalloc does not overwrite the variable binding anymore, but binds an the reigster to an additional parameter (e.g. the return value)
plan_rich
- [pypy-commit] pypy s390x-backend: prevent the base loc register to be in pool
plan_rich
- [pypy-commit] pypy s390x-backend: push pop from jitframe optimization (using store/load multiple instruction on s390x) + test
plan_rich
- [pypy-commit] pypy s390x-backend: rbigint frombytes/tobytes is specific to little endian
plan_rich
- [pypy-commit] pypy s390x-backend: realloc frame needs new size in the right register, another zll_stress test passing
plan_rich
- [pypy-commit] pypy s390x-backend: removed a debug test case from yesterdays bughunt
plan_rich
- [pypy-commit] pypy s390x-backend: removed debugging lines scattered along several places outside of the zarch backend dir
plan_rich
- [pypy-commit] pypy s390x-backend: removed debug print statement, return value of closure changed to long. a closure return value must fill a full ffi_arg
plan_rich
- [pypy-commit] pypy s390x-backend: removed debug statement, r2 is the return register on the s390x not r3
plan_rich
- [pypy-commit] pypy s390x-backend: removed debug statements, switched arguments while calling frame realloc and added a test for the pool
plan_rich
- [pypy-commit] pypy s390x-backend: removed double stack frame allocation from stacklet, zpickle requiring them pass
plan_rich
- [pypy-commit] pypy s390x-backend: removed lldebug properties from test translation (left them for debugging), fixed a bug when Const arg is provided to force_allocate_reg, it could have consumed on of the forbidden_vars
plan_rich
- [pypy-commit] pypy s390x-backend: removed longlong from s390x from cpu
plan_rich
- [pypy-commit] pypy s390x-backend: removed print statement, ztranslation_basic is now passing
plan_rich
- [pypy-commit] pypy s390x-backend: removed print statement and exchanged it with llop.debug_print (thx mjacob)
plan_rich
- [pypy-commit] pypy s390x-backend: removed print statements
plan_rich
- [pypy-commit] pypy s390x-backend: removed print statements, typo, removed push floats from stacklet because there is no need to save them
plan_rich
- [pypy-commit] pypy s390x-backend: removed size info from call builder when assembling call_assembler (it is not needed),
plan_rich
- [pypy-commit] pypy s390x-backend: removed unused import added earlier
plan_rich
- [pypy-commit] pypy s390x-backend: renamed "reg" to "r"
plan_rich
- [pypy-commit] pypy s390x-backend: renamed method on literal pool object, changed call sites to use the right function name
plan_rich
- [pypy-commit] pypy s390x-backend: replaced load 32bit imm with load imm that can handle 64 bit too
plan_rich
- [pypy-commit] pypy s390x-backend: reverted debug lines
plan_rich
- [pypy-commit] pypy s390x-backend: reverted rlib/constant.py
plan_rich
- [pypy-commit] pypy s390x-backend: rewritten many calls to use one stack frame less
plan_rich
- [pypy-commit] pypy s390x-backend: rewritten the last step of force_allocate_register, it could have taken (odd, even) instead of (even, odd)
plan_rich
- [pypy-commit] pypy s390x-backend: seems like i got rid of this nasty SIGFPE (fix point overflow), the register was dirty which could have lead to accidentally setting the FPE bit in PSW
plan_rich
- [pypy-commit] pypy s390x-backend: simplified guard_class
plan_rich
- [pypy-commit] pypy s390x-backend: some more translation issues solved
plan_rich
- [pypy-commit] pypy s390x-backend: some translation issues
plan_rich
- [pypy-commit] pypy s390x-backend: splitting tests in zarch backend, should speed up things quite a bit
plan_rich
- [pypy-commit] pypy s390x-backend: still searching for this weird problem. seems to be introduced when switching from a guard to a bridge in a very rare case
plan_rich
- [pypy-commit] pypy s390x-backend: the same for the UTF32 test
plan_rich
- [pypy-commit] pypy s390x-backend: translation changes to the instruction builder, this was up to now quite dynamic
plan_rich
- [pypy-commit] pypy s390x-backend: translation issue
plan_rich
- [pypy-commit] pypy s390x-backend: translation issue in guard_subclass, fixed autoencoding tests (had too generic argtype annotations)
plan_rich
- [pypy-commit] pypy s390x-backend: translation issues mostly and added functions and missing implementations. zrpy_releasegil translates
plan_rich
- [pypy-commit] pypy s390x-backend: two more dtype tests fixed (endian issues)
plan_rich
- [pypy-commit] pypy s390x-backend: unicode endian test issue in pypy module
plan_rich
- [pypy-commit] pypy s390x-backend: wow, how could frame regalloc even work? passing arguments in the right registers now
plan_rich
- [pypy-commit] pypy SomeRange: Create SomeRange class
rlamy
- [pypy-commit] pypy SomeRange: Inline bk.getlistdef()
rlamy
- [pypy-commit] pypy SomeRange: Make range() lists immutable
rlamy
- [pypy-commit] pypy SomeRange: Merge rpython.rtyper.lltypesystem.rrange into rpython.rtyper.rrange
rlamy
- [pypy-commit] pypy SomeRange: Separate bk.newrange() from bk.newlist()
rlamy
- [pypy-commit] pypy stmgc-c8: clarification
Raemi
- [pypy-commit] pypy stmgc-c8: possible improvement
Raemi
- [pypy-commit] pypy stmgc-c8: remove dummy allocation
Raemi
- [pypy-commit] pypy typed-cells: merge default
cfbolz
- [pypy-commit] pypy typed-cells: optimization: _pure_read already calls _read_cell, so no need to do it again
cfbolz
- [pypy-commit] pypy typed-cells: slightly different approach for integrating mutable cells and immutability:
cfbolz
- [pypy-commit] pypy value-profiling: add some tests about the new instance features
cfbolz
- [pypy-commit] pypy value-profiling: another one, for loop constants
cfbolz
- [pypy-commit] pypy value-profiling: a test for lists of known types
cfbolz
- [pypy-commit] pypy value-profiling: document the profiler
cfbolz
- [pypy-commit] pypy value-profiling: do the same thing for write
cfbolz
- [pypy-commit] pypy value-profiling: fix interaction of known class lists and cpyext
cfbolz
- [pypy-commit] pypy value-profiling: fix test_pypy_c tests
cfbolz
- [pypy-commit] pypy value-profiling: fix the interaction between record_exact_class and the heapcache
cfbolz
- [pypy-commit] pypy value-profiling: fix translation by killing dead import
cfbolz
- [pypy-commit] pypy value-profiling: integrate value-profiling and type cells better
cfbolz
- [pypy-commit] pypy value-profiling: merge default
cfbolz
- [pypy-commit] pypy value-profiling: merge default
cfbolz
- [pypy-commit] pypy value-profiling: move the valueprof to rlib
cfbolz
- [pypy-commit] pypy value-profiling: proper integration between a field being known an instance of W_IntObject or
cfbolz
- [pypy-commit] pypy value-profiling: rename valueprof to heapprof
cfbolz
- [pypy-commit] pypy value-profiling: test and fix for write_necessary logic when mixing ints and objects
cfbolz
- [pypy-commit] pypy value-profiling: test that found the problem of the previous commit
cfbolz
- [pypy-commit] pypy value-profiling: test write_necessary and make see_write return nothing
cfbolz
- [pypy-commit] pypy value-profiling: what a stupid bug
cfbolz
- [pypy-commit] pypy vmprof-newstack: add a dummy write_all_objects
fijal
- [pypy-commit] pypy vmprof-newstack: cargo-culting - maybe we should load esi first
fijal
- [pypy-commit] pypy vmprof-newstack: change pdb to some asserts
fijal
- [pypy-commit] pypy vmprof-newstack: cleanups - kill dead code
fijal
- [pypy-commit] pypy vmprof-newstack: don't release GIL here
fijal
- [pypy-commit] pypy vmprof-newstack: fix
arigo
- [pypy-commit] pypy vmprof-newstack: fixes
fijal
- [pypy-commit] pypy vmprof-newstack: fixes
fijal
- [pypy-commit] pypy vmprof-newstack: fix fix fix
fijal
- [pypy-commit] pypy vmprof-newstack: Fix for some tests: don't use .offset directly, call .getoffset()
arigo
- [pypy-commit] pypy vmprof-newstack: Fix massive leak :-)
arigo
- [pypy-commit] pypy vmprof-newstack: fix some tests
arigo
- [pypy-commit] pypy vmprof-newstack: fix tests
arigo
- [pypy-commit] pypy vmprof-newstack: fix that test
fijal
- [pypy-commit] pypy vmprof-newstack: Fix the jit to use thread-locals
arigo
- [pypy-commit] pypy vmprof-newstack: fix the test
fijal
- [pypy-commit] pypy vmprof-newstack: fix translation
fijal
- [pypy-commit] pypy vmprof-newstack: I'll never get it right
fijal
- [pypy-commit] pypy vmprof-newstack: import os x related hacks for threads & vmprof
fijal
- [pypy-commit] pypy vmprof-newstack: improve the test and make sure we emit the footer
fijal
- [pypy-commit] pypy vmprof-newstack: kill #if 0ed code
fijal
- [pypy-commit] pypy vmprof-newstack: kill more unused code
fijal
- [pypy-commit] pypy vmprof-newstack: kill some irrelevant parts
fijal
- [pypy-commit] pypy vmprof-newstack: merge
fijal
- [pypy-commit] pypy vmprof-newstack: merge default
fijal
- [pypy-commit] pypy vmprof-newstack: missing fixes for pypy
fijal
- [pypy-commit] pypy vmprof-newstack: oups
arigo
- [pypy-commit] pypy vmprof-newstack: small fixes
fijal
- [pypy-commit] pypy vmprof-newstack: some progress on the JIT front - it's still unfinished, but we have a working test
fijal
- [pypy-commit] pypy vmprof-newstack: start working on the JIT support
fijal
- [pypy-commit] pypy vmprof-newstack: ups
fijal
- [pypy-commit] pypy vmprof-newstack: ups a better way
fijal
- [pypy-commit] pypy vmprof-newstack: Use a thread-local. Kill a lot of code no longer needed.
arigo
- [pypy-commit] pypy vmprof-newstack: Use LEA, as discussed on irc
arigo
- [pypy-commit] pypy vmprof-newstack: we can have is_recursive but no get_unique_id
fijal
- [pypy-commit] pypy vmprof-newstack: Write a warning if _continuation is imported when we call _vmprof.enable()
arigo
Last message date:
Sun Jan 31 18:42:22 EST 2016
Archived on: Sun Jan 31 18:42:27 EST 2016
This archive was generated by
Pipermail 0.09 (Mailman edition).