June 2015 Archives by author
Starting: Mon Jun 1 00:33:08 CEST 2015
Ending: Tue Jun 30 19:00:13 CEST 2015
Messages: 702
- [pypy-commit] pypy issue2062: Fixed #2062 -- Treated date/datetime/time/timedelta repr like on CPython
MarkusH
- [pypy-commit] pypy issue2062: Cleaned up datetime __repr__ tests
MarkusH
- [pypy-commit] stmgc c8-locking: update todo list
Raemi
- [pypy-commit] stmgc default: Merge with c8-locking
Raemi
- [pypy-commit] pypy.org extradoc: update the python version we implement
alex_gaynor
- [pypy-commit] pypy vendor/stdlib: Upgrade 2.7 stdlib to 2.7.10
amauryfa
- [pypy-commit] pypy stdlib-2.7.10: Merge vendor/stdlib: 2.7.10
amauryfa
- [pypy-commit] pypy stdlib-2.7.10: Fix "import ssl"
amauryfa
- [pypy-commit] pypy stdlib-2.7.10: Add support for "alpn" in SSL.
amauryfa
- [pypy-commit] pypy stdlib-2.7.10: Close branch about to be merged
amauryfa
- [pypy-commit] pypy default: hg merge stdlib-2.7.10
amauryfa
- [pypy-commit] pypy default: Document merged branch
amauryfa
- [pypy-commit] pypy default: Update version numbers as indicated in doc/how-to-release.rst
amauryfa
- [pypy-commit] pypy default: Add missing test, the test suite should start now.
amauryfa
- [pypy-commit] pypy default: Fix some segfaults after a file is detached.
amauryfa
- [pypy-commit] pypy default: Avoid a crash when TextIOWrapper is not fully initialized
amauryfa
- [pypy-commit] pypy default: Fix imports in test_bz2
amauryfa
- [pypy-commit] pypy default: It is TypeError.
amauryfa
- [pypy-commit] pypy default: Add TextIOBase.newlines=None, requested in test_idle
amauryfa
- [pypy-commit] pypy default: Another method which was not exposed.
amauryfa
- [pypy-commit] pypy default: Raises ValueError on os.major(-1)
amauryfa
- [pypy-commit] pypy default: Add old buffer interface to memoryview, this is needed by the struct module.
amauryfa
- [pypy-commit] pypy default: Blindly copy a CPython change:
amauryfa
- [pypy-commit] pypy default: Fix renamed import
amauryfa
- [pypy-commit] pypy default: Fix broken translation
amauryfa
- [pypy-commit] pypy default: Another import issue
amauryfa
- [pypy-commit] pypy default: Revert the changes in memoryview.
amauryfa
- [pypy-commit] pypy py3.3: Fix sys.thread_info
amauryfa
- [pypy-commit] pypy default: More operations are allowed when a stream is detached.
amauryfa
- [pypy-commit] pypy default: CPython issue #5700: flush() was not called in close() if closefd=False.
amauryfa
- [pypy-commit] pypy default: CPython Issue #21802: The reader in BufferedRWPair now is closed even when closing
amauryfa
- [pypy-commit] pypy default: multiprocessing: Move the imports to module initialization,
amauryfa
- [pypy-commit] pypy default: Cffi Tweaks in tklib_build until we get access to TK_HEX_VERSION,
amauryfa
- [pypy-commit] pypy default: Better add '..', for the case when dirname() is empty.
amauryfa
- [pypy-commit] pypy default: _tkinter: "const" some struct members to silence gcc warnings.
amauryfa
- [pypy-commit] pypy default: Use verify(), it's simpler because it also imports the library without having to give it a real name.
amauryfa
- [pypy-commit] pypy default: Tkinter: Add support for bignum numbers.
amauryfa
- [pypy-commit] pypy default: Fix one boolean test
amauryfa
- [pypy-commit] pypy default: Another fix for boolean values
amauryfa
- [pypy-commit] pypy default: tkinter: add "WideInt" type.
amauryfa
- [pypy-commit] pypy default: tcl: Conversion from long to Bignum.
amauryfa
- [pypy-commit] pypy default: Fix test_tcl on 32bit machines.
amauryfa
- [pypy-commit] pypy default: Attempt to fix test_tcl on 32bit platforms
amauryfa
- [pypy-commit] pypy more-rposix: jit.dont_look_inside(forkpty)
amauryfa
- [pypy-commit] pypy more-rposix: Another function to hide from the jit.
amauryfa
- [pypy-commit] pypy py3.3: hg merge more-rposix, directly in py3.3 branch.
amauryfa
- [pypy-commit] pypy py3.3: oops, fix the merge
amauryfa
- [pypy-commit] pypy py3.3: Add more names to os._have_functions.
amauryfa
- [pypy-commit] pypy py3.3: os.stat() now accepts fd instead of the path
amauryfa
- [pypy-commit] pypy py3.3: Fix translation, and add a message to an assertion I don't undestand
amauryfa
- [pypy-commit] pypy py3.3: More logs
amauryfa
- [pypy-commit] pypy py3.3: Propagage no_nul when merging SomeUnicodeStrings
amauryfa
- [pypy-commit] pypy py3k: Fix syntax error
amauryfa
- [pypy-commit] cffi default: Tests for ffi.dlopen(None)
arigo
- [pypy-commit] cffi default: Issue #203: add keyword arguments to ffi.string(), ffi.buffer(),
arigo
- [pypy-commit] cffi default: Windows test fix, and documentation about ffi.dlopen()'s restriction
arigo
- [pypy-commit] cffi default: Update whatsnew
arigo
- [pypy-commit] pypy.org extradoc: Put a link to vmprof.
arigo
- [pypy-commit] stmgc c8-locking: hg merge default
arigo
- [pypy-commit] stmgc c8-locking: Turn 'modification_lock' into one real POSIX read-write lock per
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc/f0d995d5609d, branch c8-locking
arigo
- [pypy-commit] pypy default: Small tweaks
arigo
- [pypy-commit] pypy stmgc-c8: hg merge stmgc-c7
arigo
- [pypy-commit] stmgc default: hg merge c8-locking
arigo
- [pypy-commit] cffi default: Issue #205: Python 2.6 compat: "import" accept files that don't finish
arigo
- [pypy-commit] cffi default: Move these two lines outside the "with"
arigo
- [pypy-commit] cffi default: Issue #204: Spinkle "--no-user-cfg" in this test to avoid getting
arigo
- [pypy-commit] cffi default: Backed out changeset 3b9f80a475ee: not supported on python 2.6
arigo
- [pypy-commit] cffi default: Issue #204: second try
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: Update the link to Ubuntu PyPy --- go to the search page, which lists
arigo
- [pypy-commit] pypy default: Mention the modern multiple-list-types-friendly variant and not at all
arigo
- [pypy-commit] pypy default: Quickly mention rsre, and note that the `parsing' module's regexps are
arigo
- [pypy-commit] cffi default: Test and fix for ffi.verify() only on PyPy (thanks Alex Gaynor for
arigo
- [pypy-commit] cffi default: In the type parser, escape error messages and don't display the input
arigo
- [pypy-commit] pypy default: In the type parser, escape error messages and don't display the
arigo
- [pypy-commit] cffi default: Document that this example is also possible in the out-of-line mode
arigo
- [pypy-commit] stmgc c8-gil-like: A branch to have "GIL-like" behavior for inevitable transactions: one
arigo
- [pypy-commit] stmgc c8-gil-like: Plan
arigo
- [pypy-commit] cffi default: ffi.dlopen(None) does not work on Windows, and ffi.dlopen("foo") no
arigo
- [pypy-commit] cffi default: Rewords
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: A branch to simplify stmgcintf.* and see exactly what kind of interface
arigo
- [pypy-commit] pypy default: Use .decode("latin1") instead of .decode("ascii") here, which is safer
arigo
- [pypy-commit] pypy release-2.6.x: Small tweaks
arigo
- [pypy-commit] stmgc c8-gil-like: More thoughts
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy default: Redo 5cf9f578ca18 (which I only noticed by luck was reverted...).
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy default: Add and use a helper module that delays recursive calls, to turn them
arigo
- [pypy-commit] pypy default: Turn this class into a thread-local storage, to fix issues in case
arigo
- [pypy-commit] pypy default: merge heads
arigo
- [pypy-commit] pypy optresult: Add and use a helper module that delays recursive calls, to turn them
arigo
- [pypy-commit] pypy optresult: Turn this class into a thread-local storage, to fix issues in case
arigo
- [pypy-commit] pypy default: Add a test similar to test_rclass.test_flatten_convert_const, passes
arigo
- [pypy-commit] stmgc c8-gil-like: typos
arigo
- [pypy-commit] cffi default: Add comments in the C source that will hopefully be printed by the C
arigo
- [pypy-commit] cffi default: Fix the logic to handle ffi.gc(x) being called several times with equal
arigo
- [pypy-commit] cffi default: whatsnew: ffi.gc fix
arigo
- [pypy-commit] pypy default: import cffi/0a6cf359cc4a
arigo
- [pypy-commit] stmgc default: Remove 'associated_segment_num' and keep only
arigo
- [pypy-commit] cffi release-1.1: hg merge default
arigo
- [pypy-commit] cffi release-1.1: Bump version number to 1.1.1
arigo
- [pypy-commit] cffi release-1.1: Some details
arigo
- [pypy-commit] cffi release-1.1: md5/sha1
arigo
- [pypy-commit] cffi default: hg merge release-1.1
arigo
- [pypy-commit] cffi default: Support "[][...]", "[5][...]", etc.
arigo
- [pypy-commit] cffi default: Rephrase
arigo
- [pypy-commit] cffi default: Move this to the future cffi 1.2, and explain it more in cdef.rst.
arigo
- [pypy-commit] pypy default: forgot to check this in
arigo
- [pypy-commit] pypy default: merge heads
arigo
- [pypy-commit] pypy default: Issue #2060: some attempt at getting the no-sse2 version working
arigo
- [pypy-commit] pypy default: Test fix: this test used to randomly close file descriptor 1!
arigo
- [pypy-commit] pypy default: A bit more care about __del__ on half-initialized instances
arigo
- [pypy-commit] cffi default: Comment out that line from conf.py. It gives a warning with the newer
arigo
- [pypy-commit] pypy default: From this last-level helper, don't propagate further OSErrors
arigo
- [pypy-commit] pypy default: Merged in MarkusH/pypy/issue2062 (pull request #325)
arigo
- [pypy-commit] pypy default: Argh :-/
arigo
- [pypy-commit] pypy default: More test fix
arigo
- [pypy-commit] pypy default: Bump the disable_unrolling threshold in tests, because a few of them
arigo
- [pypy-commit] pypy default: Remove some very old code that is not useful any more (at least I
arigo
- [pypy-commit] pypy default: Document who is using this case
arigo
- [pypy-commit] cffi default: Issue #206: the rewrite of gc_weakref was not multithread-safe
arigo
- [pypy-commit] cffi release-1.1: Issue #206: the rewrite of gc_weakref was not multithread-safe
arigo
- [pypy-commit] cffi release-1.1: Whatsnew
arigo
- [pypy-commit] cffi release-1.1: Bump version to 1.1.2
arigo
- [pypy-commit] cffi release-1.1: md5/sha1
arigo
- [pypy-commit] cffi default: hg merge release-1.1
arigo
- [pypy-commit] pypy default: Update to cffi 1.1.2
arigo
- [pypy-commit] cffi default: Don't need an integer, actually, just object() is fine
arigo
- [pypy-commit] pypy default: Fix the ".pypy-26.so" extension, i.e. stop updating it to ".pypy-XY.so".
arigo
- [pypy-commit] stmgc c8-gil-like: hg merge default
arigo
- [pypy-commit] stmgc c8-gil-like: in-progress
arigo
- [pypy-commit] stmgc c8-gil-like: tweaks, some tests pass again
arigo
- [pypy-commit] cffi default: Add some checks that will fail in PyPy
arigo
- [pypy-commit] cffi default: Make the test more precise
arigo
- [pypy-commit] pypy default: Import cffi's 713b2badd33b and fix
arigo
- [pypy-commit] cffi default: Whatsnew
arigo
- [pypy-commit] pypy default: Don't put the "-g" option in the "gcc" invocation if we invoke it on a
arigo
- [pypy-commit] pypy default: Resolved the confusion with another confusion (but less so)
arigo
- [pypy-commit] stmgc c8-gil-like: fix fix fix
arigo
- [pypy-commit] stmgc c8-gil-like: intermediate check-in, will probably be changed again
arigo
- [pypy-commit] stmgc c8-gil-like: progress
arigo
- [pypy-commit] stmgc c8-gil-like: Essential fix here: while waiting for the inevitable transaction to
arigo
- [pypy-commit] stmgc c8-gil-like: in-progress
arigo
- [pypy-commit] stmgc c8-gil-like: fix
arigo
- [pypy-commit] stmgc c8-gil-like: fix test
arigo
- [pypy-commit] stmgc c8-gil-like: fixfix
arigo
- [pypy-commit] stmgc c8-gil-like: in-progress
arigo
- [pypy-commit] stmgc c8-gil-like: demo_random.c starts to work
arigo
- [pypy-commit] stmgc c8-gil-like: Tweaks: remove atomic_exchange() again, and use the regular
arigo
- [pypy-commit] stmgc c8-gil-like: fix
arigo
- [pypy-commit] stmgc c8-gil-like: These two disabled cases work too
arigo
- [pypy-commit] pypy default: Issue #2060: one more case
arigo
- [pypy-commit] stmgc c8-gil-like: Really remove stm_become_globally_unique_transaction and adapt
arigo
- [pypy-commit] stmgc c8-gil-like: fix
arigo
- [pypy-commit] stmgc c8-gil-like: Port demo_random2. Apparently fix things by removing this hack. I'm
arigo
- [pypy-commit] stmgc c8-gil-like: Next crash: if I manually add usleep(100) here, boom
arigo
- [pypy-commit] stmgc c8-gil-like: improve
arigo
- [pypy-commit] stmgc c8-gil-like: Tentative fix: remove the two places setting 'TS_NONE' without the
arigo
- [pypy-commit] stmgc c8-gil-like: Add the logic here, but commented out for now
arigo
- [pypy-commit] stmgc c8-gil-like: Back out f97d3ff65683: we're in a synchronize_all_threads() section,
arigo
- [pypy-commit] stmgc c8-gil-like: Fix test_extra
arigo
- [pypy-commit] stmgc c8-gil-like: Save and restore errno at the critical point
arigo
- [pypy-commit] stmgc c8-gil-like: Do a "full" implementation of stm_should_break_transaction(), as opposed
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: import stmgc, branch c8-gil-like
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: Yay, the first ztranslation test passes
arigo
- [pypy-commit] pypy optresult: Add an assert
arigo
- [pypy-commit] pypy optresult: merge heads
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: fix test
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: Fix tests
arigo
- [pypy-commit] cffi default: on pypy, the _cffi_backend module doesn't have a __file__ at all
arigo
- [pypy-commit] pypy default: import cffi/7cd6732be4e5
arigo
- [pypy-commit] pypy default: Arguably buggy corner cases of str.startswith() vs unicode.startswith()
arigo
- [pypy-commit] pypy default: Run the tests here with "--jit disable_unrolling=9999"
arigo
- [pypy-commit] pypy default: Forgot this
arigo
- [pypy-commit] stmgc c8-gil-like: Minor tweaks to make life easier for the pypy jit
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: in-progress: pypy jit
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: import stmgc/8009f12c327b
arigo
- [pypy-commit] stmgc c8-gil-like: Kill 'self' again, found a better way anyway
arigo
- [pypy-commit] stmgc c8-gil-like: Backed out changeset a8a04abfa22e
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: Fix _generate_cmp_break_transaction
arigo
- [pypy-commit] stmgc c8-gil-like: compiling with _STM_NURSERY_ZEROED=1 in non-debug mode would not
arigo
- [pypy-commit] stmgc c8-gil-like: Finish porting duhton-c8
arigo
- [pypy-commit] stmgc default: compiling with _STM_NURSERY_ZEROED=1 in non-debug mode would not
arigo
- [pypy-commit] stmgc default: improve
arigo
- [pypy-commit] stmgc c8-gil-like: Why build duhton-c8 with zeroed nursery?
arigo
- [pypy-commit] stmgc c8-gil-like: Close ready-to-merge branch
arigo
- [pypy-commit] stmgc default: hg merge c8-gil-like
arigo
- [pypy-commit] stmgc default: Mostly untested: add here the support for atomic transactions
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: import stmgc/9c72d7f52305
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: Use the new built-in atomic support
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: Bump the maximum memory from 2.5 to 7.5 GB
arigo
- [pypy-commit] stmgc default: This should be a no-op unless I'm missing something.
arigo
- [pypy-commit] pypy stmgc-c8-gil-like: import stmgc/7a87c63be4d2. adapt the JIT
arigo
- [pypy-commit] cffi default: Explicitly complain if we find 'typedef int... t; ' in a call to verify()
arigo
- [pypy-commit] stmgc queue: Pushing roots around stm_queue_put was needed only when was had
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc branch 'queue', and add rlib support for stm queues
arigo
- [pypy-commit] stmgc queue: - fix timespec_delay() to accept values of 'incr' larger than 1.0
arigo
- [pypy-commit] stmgc queue: Ah, this loop should never run more than once even now (re-add asserts)
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc/7592a0f11ac2
arigo
- [pypy-commit] pypy stmgc-c8: can't use queue.get(block=True) in an atomic context
arigo
- [pypy-commit] stmgc queue: task_done() and join() interface of Queue.Queue
arigo
- [pypy-commit] pypy stmgc-c8: add task_done() and join() to rstm's queues
arigo
- [pypy-commit] pypy stmgc-c8: task_done(), join() on stm.queue
arigo
- [pypy-commit] pypy default: issue #2067: fix the error message
arigo
- [pypy-commit] pypy stmgc-c8: translation fix
arigo
- [pypy-commit] pypy default: Don't give "libpypy-c.so" in the example programs, as this seems to
arigo
- [pypy-commit] stmgc queue: Clarify the return value of join()
arigo
- [pypy-commit] stmgc default: tweaks
arigo
- [pypy-commit] stmgc default: Tweak tweak tweak: a deadlock was possible there
arigo
- [pypy-commit] stmgc default: After cond_wait_timeout() returns with "timeout", we need to check
arigo
- [pypy-commit] stmgc queue: hg merge default
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc branch 'queue'
arigo
- [pypy-commit] pypy stmgc-c8: in-porgress: massively kill stuff in transaction.py thanks to the new queue
arigo
- [pypy-commit] stmgc default: Next version of gdb breaks existing scripts, like usual
arigo
- [pypy-commit] stmgc queue: hg merge default
arigo
- [pypy-commit] stmgc queue: Use a regular lock instead of compare-and-swap for 'old_entries'.
arigo
- [pypy-commit] stmgc queue: Missing push/pop root
arigo
- [pypy-commit] stmgc queue: tweaks
arigo
- [pypy-commit] stmgc queue: Aaaaaaaa ok found the next bug
arigo
- [pypy-commit] stmgc queue: fix fix fix minor collections
arigo
- [pypy-commit] stmgc queue: tweaks
arigo
- [pypy-commit] pypy cffi-callback-onerror: A branch to try out the cffi issue #152
arigo
- [pypy-commit] pypy cffi-callback-onerror: add the "onerror" argument to ffi.callback(), at least the out-of-line one
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc/277dd2ad5226 and fix test_transaction. Now it seems to work
arigo
- [pypy-commit] stmgc queue: oups
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc/0ef378d2da3d
arigo
- [pypy-commit] stmgc default: fixed one TODO, added another
arigo
- [pypy-commit] stmgc default: fix: must deactivate queues earlier, else major collection will try to look
arigo
- [pypy-commit] pypy stmgc-c8: update to stmgc/9ffba4fe03df and tweak the stmlog
arigo
- [pypy-commit] pypy stmgc-c8: Keep the threads around for the next call to run()
arigo
- [pypy-commit] pypy default: Test and fix for a better error message on failing imports
arigo
- [pypy-commit] pypy stmgc-c8: in-progress
arigo
- [pypy-commit] pypy stmgc-c8: merge heads
arigo
- [pypy-commit] pypy default: issue #2064: trying to call socket.close() on all sockets at exit, on Windows
arigo
- [pypy-commit] pypy default: fix test_whatsnew
arigo
- [pypy-commit] pypy default: fix this test
arigo
- [pypy-commit] pypy default: fix this test
arigo
- [pypy-commit] pypy default: Randomly try to fix test_io
arigo
- [pypy-commit] pypy default: import the cffi tests
arigo
- [pypy-commit] pypy default: Test fix
arigo
- [pypy-commit] pypy default: Duplicate 9a1683dd96e2 here
arigo
- [pypy-commit] pypy stmgc-c8: Allow 'with stm_ignored' to disable locally the insertion of
arigo
- [pypy-commit] pypy default: Simplify code
arigo
- [pypy-commit] pypy stmgc-c8: Try to add 'unsafe_write_int32' to experiment with it
arigo
- [pypy-commit] pypy stmgc-c8: translation fix
arigo
- [pypy-commit] pypy default: Never write a date in the docs without specifying the year
arigo
- [pypy-commit] pypy default: Actually, kill these paragraphs and replace them with "--shared is the
arigo
- [pypy-commit] pypy stmgc-c8: Use 'stm_dont_track_raw_accesses' instead of 'stm_ignore' here, which is
arigo
- [pypy-commit] pypy stmgc-c8: hg merge release-2.6.x
arigo
- [pypy-commit] pypy stmgc-c8: fixes for the merge
arigo
- [pypy-commit] pypy default: Issue #2069: comments and one essential fix. On one of the call paths,
arigo
- [pypy-commit] cffi default: Add a passing test
arigo
- [pypy-commit] pypy default: Mostly no-op reorganization. It avoids promoting the whole W_FunctionWrapper
arigo
- [pypy-commit] pypy default: Test and fix: ffi.addressof(lib, 'func-with-struct-args') would return
arigo
- [pypy-commit] pypy default: Give up, copy the whole test here and fix it
arigo
- [pypy-commit] pypy stmgc-c8: Add _weakref.weakkeyiddict(), a generally useful weak-keyed-dict with
arigo
- [pypy-commit] pypy stmgc-c8: Change the implementation method of threadlocalproperty().
arigo
- [pypy-commit] pypy stmgc-c8: Support the common primitive types in unsafe_write()
arigo
- [pypy-commit] pypy default: Test and fix for 'math.fsum([nan])'. It's a typo when copying from CPython's mathmodule.c.
arigo
- [pypy-commit] pypy stmgc-c8: With stm, disable this assert, which fails for now (we need to fix it
arigo
- [pypy-commit] pypy stmgc-c8: More
arigo
- [pypy-commit] pypy stmgc-c8: bah
arigo
- [pypy-commit] pypy stmgc-c8: Be more explicit in this test: it avoids the case where we add some
arigo
- [pypy-commit] pypy stmgc-c8: Fixes (sometimes of the code, sometimes of the test)
arigo
- [pypy-commit] pypy stmgc-c8: Add missing operations
arigo
- [pypy-commit] pypy stmgc-c8: updates
arigo
- [pypy-commit] pypy stmgc-c8: fix test
arigo
- [pypy-commit] pypy stmgc-c8: Fix and de-duplicate the test
arigo
- [pypy-commit] pypy default: Double mention of DEBUG_MERGE_POINT
arigo
- [pypy-commit] cffi default: Fix tests for Python 3
arigo
- [pypy-commit] pypy stmgc-c8: Forgot that arrays can also be 'stm_dont_track_raw_accesses'
arigo
- [pypy-commit] pypy stmgc-c8: Add unsafe_read(), for a different use case: multiple threads all
arigo
- [pypy-commit] pypy stmgc-c8: Fix unexpected becomes-inevitable from some math functions
arigo
- [pypy-commit] stmgc default: Try to systematically emit "wait" events when we wait
arigo
- [pypy-commit] stmgc default: Try to be a bit more careful in prof.c
arigo
- [pypy-commit] pypy stmgc-c8: import stmgc/c8ccc22dbf16
arigo
- [pypy-commit] pypy stmgc-c8: Update print_stm_log
arigo
- [pypy-commit] cffi default: Clarify the weakdict example
arigo
- [pypy-commit] stmgc default: - use CLOCK_MONOTONIC_RAW instead of CLOCK_MONOTONIC, because the latter is not
arigo
- [pypy-commit] pypy stmgc-c8: more tests
arigo
- [pypy-commit] pypy stmgc-c8: In the JIT, copy the stm_dont_track_raw_accesses logic for getfield_raw
arigo
- [pypy-commit] pypy stmgc-c8: update to stmgc/9f966d34d3be
arigo
- [pypy-commit] pypy stmgc-c8: Progress, but still something TO DO
arigo
- [pypy-commit] pypy default: Plug the logic from find_initializing_stores() to also detect and remove
arigo
- [pypy-commit] pypy stmgc-c8: Plug the logic from find_initializing_stores() to also detect and remove
arigo
- [pypy-commit] pypy stmgc-c8: Copy this logic here
arigo
- [pypy-commit] pypy default: Theoretical fix. In practice, I think it cannot really occur that a
arigo
- [pypy-commit] pypy default: merge heads
arigo
- [pypy-commit] pypy stmgc-c8: Theoretical fix. In practice, I think it cannot really occur that a
arigo
- [pypy-commit] cffi default: Test and fix: Python 3.5 (at least the current beta) is unhappy
arigo
- [pypy-commit] pypy default: cffi/9d86b893ad22: add 'lib.__name__'
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy default: Attempt to fix the test on some machines, where the open() fails with
arigo
- [pypy-commit] pypy default: Test and fix for a case of mutating kwargs dictionaries. Thanks
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy default: Document the difference with the readline module
arigo
- [pypy-commit] pypy default: grammar
arigo
- [pypy-commit] pypy default: Drop "some of us believe", as the cpython issue links to a 29c3 talk
arigo
- [pypy-commit] pypy default: Link to the 29c3 talk
arigo
- [pypy-commit] cffi default: ffi.include(ffi) doesn't make sense and hangs right now
arigo
- [pypy-commit] cffi default: Issue #207: fix for anonymous enums with ffi.include()
arigo
- [pypy-commit] pypy default: Mention issue #2030 in cpython_differences
arigo
- [pypy-commit] pypy default: Issue #209: check for dereferencing NULL pointers
arigo
- [pypy-commit] cffi default: Issue #209: check for dereferencing NULL pointers
arigo
- [pypy-commit] pypy default: Fix translation (hopefully)
arigo
- [pypy-commit] pypy vmprof-review: A branch to review vmprof
arigo
- [pypy-commit] pypy vmprof-review: an XXX for now
arigo
- [pypy-commit] pypy vmprof-review: Fix this test
arigo
- [pypy-commit] pypy default: Still trying to fix (more of the same)
arigo
- [pypy-commit] pypy vmprof-review: potential crash
arigo
- [pypy-commit] benchmarks default: add a synthetic sqlite benchmark, mostly to stress CFFI
cfbolz
- [pypy-commit] benchmarks default: ouch. sorry about that.
cfbolz
- [pypy-commit] pypy default: don't create HeapCacheValue entries just to escape them
cfbolz
- [pypy-commit] pypy default: move the check whether something is an ovf operation from
cfbolz
- [pypy-commit] pypy default: in the heapcache don't do the same dict lookups twice for every getfield and
cfbolz
- [pypy-commit] pypy optresult: merge default
cfbolz
- [pypy-commit] pypy optresult: disable the unroll-related change for now, it breaks stuff
cfbolz
- [pypy-commit] pypy optresult: typo
cfbolz
- [pypy-commit] pypy optresult: use proper new interface
cfbolz
- [pypy-commit] pypy default: implement file.readlines more efficiently by not wrapping every individual
cfbolz
- [pypy-commit] pypy optresult: add a workaround
fijal
- [pypy-commit] pypy optresult: fix RECORD_KNOWN_CLASS
fijal
- [pypy-commit] pypy optresult: minor fixes
fijal
- [pypy-commit] pypy optresult: allow None in failargs;
fijal
- [pypy-commit] pypy optresult: fix the backend
fijal
- [pypy-commit] pypy optresult: work on rpython and recursive stuff
fijal
- [pypy-commit] pypy optresult: work on rpython and recursive stuff
fijal
- [pypy-commit] pypy optresult: start working on the virtualizable
fijal
- [pypy-commit] pypy optresult: basic fixes to vstring
fijal
- [pypy-commit] pypy optresult: fixes to virtualizable
fijal
- [pypy-commit] pypy optresult: fix tmp_callback
fijal
- [pypy-commit] pypy optresult: an attempt to fix exception handling in the new model
fijal
- [pypy-commit] pypy optresult: one more thing
fijal
- [pypy-commit] pypy optresult: fix here
fijal
- [pypy-commit] pypy optresult: finish fixing exceptions
fijal
- [pypy-commit] pypy optresult: two more small fixes
fijal
- [pypy-commit] pypy optresult: fix fix fix fix
fijal
- [pypy-commit] pypy optresult: fix quasiimmut + rpython fixes
fijal
- [pypy-commit] pypy optresult: few fixes
fijal
- [pypy-commit] pypy optresult: various random rpython fixes
fijal
- [pypy-commit] pypy default: seems we fixed ll2ctypes
fijal
- [pypy-commit] pypy default: merge
fijal
- [pypy-commit] pypy optresult: start implementing virtual array of structs
fijal
- [pypy-commit] pypy optresult: fight resume of virtual arrays of structs
fijal
- [pypy-commit] pypy optresult: fix test_blackhole
fijal
- [pypy-commit] pypy optresult: minor fixes
fijal
- [pypy-commit] pypy optresult: start fighting with raw buffers
fijal
- [pypy-commit] pypy optresult: finish test_fficall
fijal
- [pypy-commit] pypy optresult: whack until most tests pass
fijal
- [pypy-commit] pypy optresult: implement more of vstring
fijal
- [pypy-commit] pypy optresult: finish vstring, I hope
fijal
- [pypy-commit] pypy optresult: small fixes
fijal
- [pypy-commit] pypy optresult: fix test_calling_convention
fijal
- [pypy-commit] pypy optresult: fragile fragile
fijal
- [pypy-commit] pypy optresult: use the proper descr
fijal
- [pypy-commit] pypy optresult: unify the return value
fijal
- [pypy-commit] pypy optresult: whack at some tests
fijal
- [pypy-commit] pypy optresult: unify the interface
fijal
- [pypy-commit] pypy optresult: those asserts are no longer valid due to parent_descr
fijal
- [pypy-commit] pypy optresult: disable those for now
fijal
- [pypy-commit] pypy optresult: fixes
fijal
- [pypy-commit] pypy optresult: support GETFIELD_RAW_R
fijal
- [pypy-commit] pypy optresult: grr
fijal
- [pypy-commit] pypy optresult: sanity check and disable gen_store_back until we test it
fijal
- [pypy-commit] pypy optresult: arbitrary fix
fijal
- [pypy-commit] pypy optresult: try to fix annotation
fijal
- [pypy-commit] pypy optresult: a test and a fix
fijal
- [pypy-commit] pypy optresult: oops
fijal
- [pypy-commit] pypy optresult: rpythonify
fijal
- [pypy-commit] pypy optresult: rpythonize this part too
fijal
- [pypy-commit] pypy optresult: disable it better
fijal
- [pypy-commit] pypy optresult: enable this check always, we can always disable it later
fijal
- [pypy-commit] pypy optresult: skip one too direct test. fix one case of virtuals
fijal
- [pypy-commit] pypy optresult: fix test_w*
fijal
- [pypy-commit] pypy optresult: blind fix
fijal
- [pypy-commit] pypy optresult: kill dead code
fijal
- [pypy-commit] pypy optresult: track the length of strings
fijal
- [pypy-commit] pypy optresult: fight with spurious test failures
fijal
- [pypy-commit] pypy optresult: fix for llgraph backend
fijal
- [pypy-commit] pypy optresult: can't do it due to raise_continue_running_normally
fijal
- [pypy-commit] pypy optresult: whack at some tests
fijal
- [pypy-commit] pypy optresult: fixes to vstring, we need unrolling otherwise it's chasing a wild goose :/
fijal
- [pypy-commit] pypy optresult: start killing things
fijal
- [pypy-commit] pypy optresult: kill kill kill
fijal
- [pypy-commit] pypy optresult: this is I think necessary
fijal
- [pypy-commit] pypy optresult: no, we already call this from force_box
fijal
- [pypy-commit] pypy optresult: start fighting with unrolling
fijal
- [pypy-commit] pypy optresult: redisable unrolling - let's get everything else going first
fijal
- [pypy-commit] pypy optresult: a test and a fix
fijal
- [pypy-commit] pypy optresult: a test, explanation and update TODO
fijal
- [pypy-commit] pypy optresult: give up on integrating rewrite_consptr. we allow stuff to be modified in place in this simple case
fijal
- [pypy-commit] pypy optresult: fix all the llsupport tests
fijal
- [pypy-commit] pypy optresult: fix fix fix
fijal
- [pypy-commit] pypy optresult: whack at test_resume until it passes
fijal
- [pypy-commit] pypy optresult: comment out unused function
fijal
- [pypy-commit] pypy optresult: fix
fijal
- [pypy-commit] pypy optresult: try to collect a few more times to see if it's a problem
fijal
- [pypy-commit] pypy optresult: did not help, revert
fijal
- [pypy-commit] pypy optresult: random fix
fijal
- [pypy-commit] pypy optresult: try to fix annotation
fijal
- [pypy-commit] pypy optresult: do it that way
fijal
- [pypy-commit] pypy optresult: a bit blindly implement direct_call_release_gil
fijal
- [pypy-commit] pypy optresult: fix
fijal
- [pypy-commit] pypy optresult: fix the spelling
fijal
- [pypy-commit] pypy optresult: missed a None here
fijal
- [pypy-commit] pypy optresult: one potential stupid segfault
fijal
- [pypy-commit] pypy optresult: an experiment in specialization
fijal
- [pypy-commit] pypy optresult: fix for running tests
fijal
- [pypy-commit] pypy optresult: oops
fijal
- [pypy-commit] pypy optresult: pfff
fijal
- [pypy-commit] pypy optresult: another oops
fijal
- [pypy-commit] pypy optresult: the specialization seems to be an overkill
fijal
- [pypy-commit] pypy optresult: an attempt to make heapcache use new mechanism of tracking stuff
fijal
- [pypy-commit] pypy optresult: of course
fijal
- [pypy-commit] pypy optresult: disable those features in favor of having a working --fork-before for now
fijal
- [pypy-commit] pypy optresult: write a comment
fijal
- [pypy-commit] pypy optresult: backout an attempt on heapcache, does not seem to help
fijal
- [pypy-commit] pypy disable-unroll-for-short-loops: an attempt to disable unrolling if we hit a certain threshold
fijal
- [pypy-commit] pypy disable-unroll-for-short-loops: try to attack the problem differently, a bit unclear how to test the beast
fijal
- [pypy-commit] pypy optresult: a failing test
fijal
- [pypy-commit] pypy optresult: tiny fix
fijal
- [pypy-commit] benchmarks single-run: another, slightly different, benchmark
fijal
- [pypy-commit] benchmarks default: add another, slightly different, benchmark
fijal
- [pypy-commit] pypy disable-unroll-for-short-loops: "tweak" the parameter
fijal
- [pypy-commit] pypy optresult: try to disable pending setfields
fijal
- [pypy-commit] pypy optresult: revert the previous checkin does not seem to be the problem
fijal
- [pypy-commit] pypy optresult: A fix for preserving pointerness
fijal
- [pypy-commit] pypy optresult: pfff
fijal
- [pypy-commit] pypy disable-unroll-for-short-loops: close to be merged branch
fijal
- [pypy-commit] pypy default: Merge disable-unroll-for-short-loops (which actually is supposed to be
fijal
- [pypy-commit] pypy optresult: This can be NULL here
fijal
- [pypy-commit] pypy optresult: disable this import, again for --fork-before
fijal
- [pypy-commit] pypy optresult: fix pypyjit.py
fijal
- [pypy-commit] pypy optresult: there is no point in trying to preserve the Const here
fijal
- [pypy-commit] pypy optresult: be more specific, I'm sure we have more of those
fijal
- [pypy-commit] pypy optresult: gah, dont release the GIL here
fijal
- [pypy-commit] pypy optresult: a hack to make sure that we have a correct type in fielddescr
fijal
- [pypy-commit] pypy optresult: check only for gc stuff
fijal
- [pypy-commit] pypy optresult: strgetitem handles self.make_equal_to already
fijal
- [pypy-commit] pypy optresult: an attempt to sort out vtable mess
fijal
- [pypy-commit] pypy optresult: add a fast path to avoid issues with JitVirtualRef
fijal
- [pypy-commit] pypy optresult: add a comment
fijal
- [pypy-commit] pypy optresult: fixes
fijal
- [pypy-commit] pypy optresult: merge
fijal
- [pypy-commit] pypy optresult-unroll: scarily enough to start passing unrolling tests
fijal
- [pypy-commit] pypy optresult-unroll: more passing tests
fijal
- [pypy-commit] pypy optresult: kill box usage
fijal
- [pypy-commit] pypy optresult: fix wrong use of check isinstance(x, AbstractResOp)
fijal
- [pypy-commit] pypy optresult-unroll: merge optresult
fijal
- [pypy-commit] pypy optresult-unroll: progress
fijal
- [pypy-commit] pypy default: improve the test
fijal
- [pypy-commit] pypy optresult-unroll: start working towards short preamble
fijal
- [pypy-commit] pypy optresult-unroll: small fix
fijal
- [pypy-commit] pypy default: merge
fijal
- [pypy-commit] pypy default: kill the alignment here
fijal
- [pypy-commit] pypy default: rename the file, it's not related to asmgcc at all
fijal
- [pypy-commit] pypy optresult-unroll: start passing construction of args
fijal
- [pypy-commit] pypy optresult-unroll: add GUARD_VALUE for short preamble
fijal
- [pypy-commit] pypy optresult-unroll: kill pdb
fijal
- [pypy-commit] pypy optresult-unroll: fix, I think
fijal
- [pypy-commit] pypy default: try to fix for support
fijal
- [pypy-commit] pypy default: port the changes from upstream
fijal
- [pypy-commit] pypy release-2.6.x: update release announcement to cffi 1.1
mattip
- [pypy-commit] pypy.org extradoc: update for 2.6.0 release
mattip
- [pypy-commit] pypy default: merge release-2.6.x into default for pypy/doc changes
mattip
- [pypy-commit] pypy default: fix merge
mattip
- [pypy-commit] jitviewer default: Added tag pypy-2.6.0 for changeset 3a0152b4ac6b
mattip
- [pypy-commit] pypy default: fix translation ("argument not constant")
mattip
- [pypy-commit] pypy default: test, fix to support creating a record array from a different record array
mattip
- [pypy-commit] pypy default: add context argument to __array_wrap__ in ufuncs
mattip
- [pypy-commit] pypy default: test, fix record array creation with int value broadcast to internal array
mattip
- [pypy-commit] pypy default: special case buffer -> uint8 dtype
mattip
- [pypy-commit] pypy default: test, fix flags.owndata
mattip
- [pypy-commit] pypy default: test, implement part of dtype(align=True), missing isalignstruct in dtype.flags
mattip
- [pypy-commit] pypy default: assign to self.flags later in __init__ and fewer times, seems to help immutablility
mattip
- [pypy-commit] pypy default: test, implement kwargs to astype
mattip
- [pypy-commit] pypy default: sanity check
mattip
- [pypy-commit] pypy default: refactor ndarray.flags in cpyext
mattip
- [pypy-commit] pypy default: allow only C, F in concrete type order
mattip
- [pypy-commit] pypy default: test, fix converting object dtype to str
mattip
- [pypy-commit] pypy dtypes-compatability: add and modify tests for numpy compatibility, fix shape failures
mattip
- [pypy-commit] pypy dtypes-compatability: fix subclass test, break translation
mattip
- [pypy-commit] pypy dtypes-compatability: add, implement flags, fix alignment
mattip
- [pypy-commit] pypy dtypes-compatability: sort out bogus tests differentiation between pypy and cpython
mattip
- [pypy-commit] pypy dtypes-compatability: fix translation
mattip
- [pypy-commit] pypy default: tagging only after buildbot runs produces a cleaner flow
mattip
- [pypy-commit] pypy release-2.6.x: tagging only after buildbot runs produces a cleaner flow
mattip
- [pypy-commit] pypy dtypes-compatability: fix z_translation, test_compile (arigato)
mattip
- [pypy-commit] pypy dtypes-compatability: add tests for metadata, dtype-from-dict
mattip
- [pypy-commit] pypy run-create_cffi_imports: start to add a build task to create cffi_imports
mattip
- [pypy-commit] benchmarks default: disable cffi-dependent benchmark until translation runs cffi-import
mattip
- [pypy-commit] pypy dtypes-compatability: test align more thoroughly (from numpy tests), percolate align argument
mattip
- [pypy-commit] pypy dtypes-compatability: wip - fixing align
mattip
- [pypy-commit] pypy run-create_cffi_imports: merge default into branch
mattip
- [pypy-commit] pypy run-create_cffi_imports: add try/except (arigato)
mattip
- [pypy-commit] pypy run-create_cffi_imports: create_cffi_import_libraries does not raise rather indicates which imports failed
mattip
- [pypy-commit] pypy run-create_cffi_imports: error message, update whatsnew
mattip
- [pypy-commit] pypy run-create_cffi_imports: close branch to be merged
mattip
- [pypy-commit] pypy default: merge run-create_cffi_imports, which builds cffi import libraries as part of translation
mattip
- [pypy-commit] benchmarks default: Backed out 45bdb6aca4bb, re-enable cffi-dependent benchmakr
mattip
- [pypy-commit] pypy default: remove prefix from MSVC exe name, since prefix was not handled in driver.compute_exe_name
mattip
- [pypy-commit] pypy default: be more consistent with win32 identification macro
mattip
- [pypy-commit] pypy pypy3-release-2.6.x: small win32 fixes
mattip
- [pypy-commit] pypy pypy3-release-2.6.x: fix annotation on narrow builds (win32)
mattip
- [pypy-commit] pypy py3k: Make curses.tigetstr/tigetnum/tigetflag handle Unicode strings
mgedmin
- [pypy-commit] pypy py3k: Fix curses text-output functions and methods
mgedmin
- [pypy-commit] pypy py3k: Link against libncursesw, for Unicode support
mgedmin
- [pypy-commit] pypy py3k: Merge with upstream
mgedmin
- [pypy-commit] pypy py3k: Adapt test to check that type.__eq__(a, b) returns the same as object.__eq__(a, b).
mjacob
- [pypy-commit] pypy py3k: hg merge default
mjacob
- [pypy-commit] pypy py3k: Use threading.Lock() instead of thread.allocate_lock().
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Remove merge conflict marker which was accidently left.
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Fix.
mjacob
- [pypy-commit] pypy py3k: Fix.
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Remove lib_pypy/_struct.py which doesn't work on py3k and is redundant anyway.
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Add missing import.
mjacob
- [pypy-commit] pypy py3k: Add a workaround for Python 2.7.6 raising a SyntaxError when importing a file that contains Py3k-compatible exec(code, d) calls in certain contexts.
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Correct indentation to be a multiple of four.
mjacob
- [pypy-commit] pypy py3k: Call dict.copy() instead of dict.items() here because dict.items() does not copy the contents on Py3k.
mjacob
- [pypy-commit] pypy py3k: Add the struct module to this test's spaceconfig because the pure Python version of it was removed on this branch.
mjacob
- [pypy-commit] pypy py3k: Make these tests, which rely on the GIL being released, more reliable.
mjacob
- [pypy-commit] pypy py3k: Let the trace module ignore some modules which get freezed at translation-time.
mjacob
- [pypy-commit] pypy py3k: Fix.
mjacob
- [pypy-commit] pypy py3k: Add the struct module to more tests' spaceconfig (the pure Python version was removed on this branch).
mjacob
- [pypy-commit] pypy py3k: Try to make tests with threads and signals more reliable.
mjacob
- [pypy-commit] pypy py3k: Add workaround for when host interpreter has slightly different errnos than interpreter under test.
mjacob
- [pypy-commit] pypy py3k: Fix test.
mjacob
- [pypy-commit] pypy py3k: Make this test, which involves threads and signals, more likely to pass.
mjacob
- [pypy-commit] pypy py3k: Expose the __class__ cell to Python code.
mjacob
- [pypy-commit] pypy py3k: bytes(obj) should call obj.__index__() instead of obj.__int__().
mjacob
- [pypy-commit] pypy py3k: Back out changeset b84f64795e2b because I referred the wrong issue in the commit message.
mjacob
- [pypy-commit] pypy py3k: bytes(obj) should call obj.__index__() instead of obj.__int__().
mjacob
- [pypy-commit] pypy py3k: Remove 'utf-8' argument from calls to str.encode() in this file because it's already the default argument.
mjacob
- [pypy-commit] pypy py3k: Add a fallback to the ncurses library if ncursesw is not available.
mjacob
- [pypy-commit] pypy py3k: Fix a bug that caused e.g. None.__eq__ to evaluate to an unbound function.
mjacob
- [pypy-commit] pypy py3k: Merged in yuyichao/pypy/py3k (pull request #326)
mjacob
- [pypy-commit] pypy py3k: Port virtualenv workaround for shared builds from the 2.7 stdlib to the 3.2 stdlib.
mjacob
- [pypy-commit] pypy py3k: Fix CFFI test.
mjacob
- [pypy-commit] pypy py3k: Fix translation.
mjacob
- [pypy-commit] pypy py3.3: hg merge py3k
mjacob
- [pypy-commit] pypy py3.3: Re-add has_so_extension() to module/imp/importing.py (it was removed during the last merge because I thought it's dead code).
mjacob
- [pypy-commit] pypy py3.3: Port _lzma to CFFI 1.0.
mjacob
- [pypy-commit] pypy py3.3: Port _decimal to CFFI 1.0.
mjacob
- [pypy-commit] pypy py3k: hg merge default
mjacob
- [pypy-commit] pypy default: hide app_main's frames. this breaks sys.exc_info but py3 offers a workaround
pjenvey
- [pypy-commit] pypy default: adapt to py2.7
pjenvey
- [pypy-commit] pypy default: add __pypy__.get_hidden_tb so pypy2's app_main can grab the current traceback
pjenvey
- [pypy-commit] pypy default: hide more init functions
pjenvey
- [pypy-commit] pypy vecopt: distinct between input/output argument in vector type conversion
plan_rich
- [pypy-commit] pypy vecopt: reverted call2, list + index left the access in the trace
plan_rich
- [pypy-commit] pypy vecopt: started to add cost model, all enabled zjit tests passing now
plan_rich
- [pypy-commit] pypy vecopt: rpython compliant, started to impl cost model
plan_rich
- [pypy-commit] pypy vecopt: costmodel impl extended
plan_rich
- [pypy-commit] pypy vecopt: activated all but 3 zjit tests (pow, take missing), all others pass
plan_rich
- [pypy-commit] pypy vecopt: removed unused dict, lead to rpy error (no values included)
plan_rich
- [pypy-commit] pypy vecopt: removed the unpacking of the values in the trace of call2
plan_rich
- [pypy-commit] pypy vecopt: excluded <OP>_GC from vectorization. Those ops can load complex objects which are not subject of this optimization
plan_rich
- [pypy-commit] pypy vecopt: wrong assumption: it can happen that a call that is exception guarded is not immediatley after the call
plan_rich
- [pypy-commit] pypy vecopt: removed primitive type mixin
plan_rich
- [pypy-commit] pypy vecopt: expanding variables/constants are not inlined before the instruction but gathered in another list to be prepended to the label operation
plan_rich
- [pypy-commit] pypy vecopt: prepending invariant operations before the label
plan_rich
- [pypy-commit] pypy vecopt: vector boxes in label and jump args are not correctly dispatched as xmm registers
plan_rich
- [pypy-commit] pypy vecopt: remember expansion (reduces register pressure if constant/variable is use more often). but not for heterogeneous expanded vectors, I don't think this happens frequently
plan_rich
- [pypy-commit] pypy vecopt: merged release-2.6.x
plan_rich
- [pypy-commit] pypy vecopt: added missing vec counters to jitprof test
plan_rich
- [pypy-commit] pypy vecopt: adapted test_zjit to changes in micro numpy
plan_rich
- [pypy-commit] pypy vecopt: int expand 64 bit working
plan_rich
- [pypy-commit] pypy vecopt: silenced rpython complaints
plan_rich
- [pypy-commit] pypy vecopt: problem with float abs (expanded variable but did not remove second use), andps constant pointed to neg constant
plan_rich
- [pypy-commit] pypy vecopt: int expansion for int16 and int8 added, int32/16 test added first already passes
plan_rich
- [pypy-commit] pypy vecopt: updating the input type. when items are packed their count changes, thus the input and output type must be updated
plan_rich
- [pypy-commit] pypy vecopt: extract instructions sse4 had some wrong parameters in assembler
plan_rich
- [pypy-commit] pypy vecopt: tested insert x86 opcodes
plan_rich
- [pypy-commit] pypy vecopt: removed manual test since it is tested in auto test now
plan_rich
- [pypy-commit] pypy vecopt: reverted the 8 immediate suffix (solved differently for tests)
plan_rich
- [pypy-commit] pypy vecopt: started to find reduce/accumulation functions that are vectorizable (e.g. sum)
plan_rich
- [pypy-commit] pypy vecopt: added accumulation pair
plan_rich
- [pypy-commit] pypy vecopt: finding reduceables works, scheduling needs to be adapted next
plan_rich
- [pypy-commit] pypy vecopt: removed allocation of comparison objects to track guard comparisons. This is handled by the guard strengthing operation
plan_rich
- [pypy-commit] pypy vecopt: moved out guard strengthening (on arith level) and scheduling from vectorize.py and schedule.py
plan_rich
- [pypy-commit] pypy vecopt: finished refactor the structure
plan_rich
- [pypy-commit] pypy vecopt: can_be_packed did not consider the case if origin pack is None, some small other refactorings
plan_rich
- [pypy-commit] pypy vecopt: scheduler emits packs that are interdependent (only within pack and marked accum)
plan_rich
- [pypy-commit] pypy vecopt: generating vector box for accumulation before the label and renaming occurances
plan_rich
- [pypy-commit] pypy vecopt: accumulation box is now used instead of the scalar box, need to adapt guard exit to produce correct result
plan_rich
- [pypy-commit] pypy vecopt: need to call method to get the result :)
plan_rich
- [pypy-commit] pypy vecopt: added test to ensure sum is not prohibited by cost calculation
plan_rich
- [pypy-commit] pypy vecopt: updated tests to provide arguments for some changed methods. float vector box signed is always false
plan_rich
- [pypy-commit] pypy vecopt: in the case #pack > 2 the accum is not split
plan_rich
- [pypy-commit] pypy vecopt: moving the type determination from the scheduling transformation to the find pack routines,
plan_rich
- [pypy-commit] pypy vecopt: works now for the int16 case
plan_rich
- [pypy-commit] pypy vecopt: preventing accumulator flush within the trace. the plan is only to support in in guard exits
plan_rich
- [pypy-commit] pypy vecopt: preventing the transformation of cumsum (some problems that followed from last changes)
plan_rich
- [pypy-commit] jitviewer vmprof-address: adding vmprof address to the right side of a loop
plan_rich
- [pypy-commit] jitviewer vmprof-address: merged defaut
plan_rich
- [pypy-commit] pypy vecopt: vector boxes need to be considered on guard exit
plan_rich
- [pypy-commit] pypy vmprof-address: changes to view the bootstrap address of vmprof
plan_rich
- [pypy-commit] pypy vmprof-address: manually loading address from log to get all addresses and be able to display all assembler dumps
plan_rich
- [pypy-commit] pypy vecopt: loading all code regions instead of the merged ones from World
plan_rich
- [pypy-commit] pypy vmprof-address: sorting the code ranges again
plan_rich
- [pypy-commit] pypy vecopt: added a new field to the resume guard descr to handle accumulation variables at guard exit
plan_rich
- [pypy-commit] pypy vecopt: signext (due to its missing x86 opcodes) is weigthed as "not a profitable operation"
plan_rich
- [pypy-commit] pypy vecopt: added tests and parameterized one
plan_rich
- [pypy-commit] pypy vecopt: added uint vector addition test
plan_rich
- [pypy-commit] pypy vecopt: removed set_trace call
plan_rich
- [pypy-commit] pypy vecopt: added llgraph impl for reduce (+,-,*)n
plan_rich
- [pypy-commit] pypy vecopt: some test changes, removed an error when the output type is none
plan_rich
- [pypy-commit] pypy vecopt: added test case to sum on llgraph
plan_rich
- [pypy-commit] pypy vecopt: error in the test case added array instead of summed scalar
plan_rich
- [pypy-commit] pypy vecopt: jit.promote in the numpy internal iterator generates slightly different traces, adjusted tests with the new constant
plan_rich
- [pypy-commit] pypy vecopt: added test case to ensure correctness of transformation and find all missing assembler implementations
plan_rich
- [pypy-commit] pypy regalloc: storing the variable use/def information in the tuple of longevity as third parameter
plan_rich
- [pypy-commit] pypy regalloc: exchanged the heuristic for the register allocator. it now takes the live range whose next use is furthest in the future
plan_rich
- [pypy-commit] pypy vecopt: remembering the position of the guard exit to resume the regallocator at a guard exit
plan_rich
- [pypy-commit] pypy vecopt: typo in arm assembler.py
plan_rich
- [pypy-commit] pypy vecopt: rpython issue
plan_rich
- [pypy-commit] pypy regalloc: added logic to put every variable into the register depending on how far away the next usage is (less distance is more likely to end up in a register)
plan_rich
- [pypy-commit] pypy regalloc: added more information to the output of regalloc.py
plan_rich
- [pypy-commit] pypy vecopt: add a new doc file to describe vecopt, describe what the current opt. is capable of and added some limitations
plan_rich
- [pypy-commit] pypy vecopt: timing the vecopt and printing it to the logfiles
plan_rich
- [pypy-commit] pypy vecopt: added some temporary timing code for the trace execution
plan_rich
- [pypy-commit] pypy vecopt: missing bracket
plan_rich
- [pypy-commit] pypy vecopt: moved around the timing to be able to translate the project with it
plan_rich
- [pypy-commit] pypy vecopt: some changes after being tested with test_zjit
plan_rich
- [pypy-commit] pypy vecopt: missing format character
plan_rich
- [pypy-commit] pypy vecopt: added test case for conversion int64 <-> int8
plan_rich
- [pypy-commit] pypy vecopt: started to bail out for some operations that cannot be implemented efficiently in SSE4.1
plan_rich
- [pypy-commit] pypy vecopt: missing variable definition
plan_rich
- [pypy-commit] pypy vecopt: setting to safe default
plan_rich
- [pypy-commit] pypy vecopt: added a new resop class for casting operations, added a test to ensure they are created correctly
plan_rich
- [pypy-commit] pypy vecopt: preventing int signext from >32 -> <32
plan_rich
- [pypy-commit] pypy vecopt: added comment in doc
plan_rich
- [pypy-commit] pypy vecopt: reformated the log output to time the vecopt traces
plan_rich
- [pypy-commit] pypy vecopt: well, should be int_mul not int_add...
plan_rich
- [pypy-commit] pypy vecopt: modified tests for missing packed int64 mul (was incorrect before) and prevented int64->int16 conversion
plan_rich
- [pypy-commit] pypy vecopt: doc additions, reenabled the int8 expand test (passes now)
plan_rich
- [pypy-commit] pypy vecopt: logging condition changed for timing in optimize_trace
plan_rich
- [pypy-commit] pypy vecopt: rpython translation issues
plan_rich
- [pypy-commit] pypy vecopt: adding guards as vector instructions. i'm not yet sure how this will work out, but could help to generate better loops for reductions
plan_rich
- [pypy-commit] pypy vecopt: docu comment
plan_rich
- [pypy-commit] pypy vecopt: moved timing functions into the execute assembler, commented out the guard as vec opt for this translation test
plan_rich
- [pypy-commit] pypy vecopt: passing all scheduling tests again
plan_rich
- [pypy-commit] pypy vecopt: don't follow wrong dependency chains, excluded fail args
plan_rich
- [pypy-commit] pypy vecopt: splitting must be done differently
plan_rich
- [pypy-commit] pypy vecopt: rewritten splitting of packs, added asserts to ensure the impl assumptions are correct. some tests broke (it is not yet finished)
plan_rich
- [pypy-commit] pypy vecopt: tyring to make things more easier, complexity gets hard to manage if extending the accumulation. trying to prevent the splitting entering the scheduling (work in progress)
plan_rich
- [pypy-commit] pypy vecopt: finishing up changes
plan_rich
- [pypy-commit] pypy vecopt: test_vectorize passing again
plan_rich
- [pypy-commit] pypy vecopt: rewritten scheduling tests to add type (could not be inferred easily), cost model passing again
plan_rich
- [pypy-commit] pypy vecopt: added guard_true/false for vector register as first argument
plan_rich
- [pypy-commit] pypy vecopt: all & any (but not any casting from float) execute correctly
plan_rich
- [pypy-commit] pypy vecopt: fixed guard_false for packed arguments, the last test (set_slice) not working
plan_rich
- [pypy-commit] pypy vecopt: pass zjit, splitting is not necessary anymore, because the packing phase determines if it is able to combine pairs or not
plan_rich
- [pypy-commit] pypy vecopt: rpython translation issues, adding reshape operation to zjit tests
plan_rich
- [pypy-commit] pypy use_min_scalar: correct handling of scalars for simple binary ufuncs
rlamy
- [pypy-commit] pypy use_min_scalar: correct handling of scalars for non-simple binary ufuncs
rlamy
- [pypy-commit] pypy use_min_scalar: Switch to the scalar fast path earlier in W_Ufunc2.call()
rlamy
- [pypy-commit] pypy use_min_scalar: use_min_scalar can never be true for unary ufuncs
rlamy
- [pypy-commit] pypy use_min_scalar: Close branch use_min_scalar
rlamy
- [pypy-commit] pypy default: Merged use_min_scalar into default
rlamy
- [pypy-commit] pypy default: update whatsnew
rlamy
- [pypy-commit] pypy unicode-dtype: Enable np.unicode_
rlamy
- [pypy-commit] pypy unicode-dtype: Reimplement W_UnicodeBox as a simple wrapper around an interp-level unicode object
rlamy
- [pypy-commit] pypy default: Move implementation of FlexibleType.to_str() to W_FlexibleBox
rlamy
- [pypy-commit] pypy default: Create W_Dtype.runpack_str() and simplify its callers
rlamy
- [pypy-commit] pypy default: Fix translation and allow order='K'
rlamy
- [pypy-commit] pypy default: Create W_Dtype.store()
rlamy
- [pypy-commit] pypy default: Create W_Dtype.read()
rlamy
- [pypy-commit] pypy default: make dtype argument to itemtype.read() mandatory
rlamy
- [pypy-commit] pypy default: Create W_Dtype.read_bool()
rlamy
- [pypy-commit] pypy default: Remove 'native' attribute from itemtypes
rlamy
- [pypy-commit] pypy default: Ensure that W_Dtype.byteorder is a char, not a string, and hopefully fix performance regression
rlamy
- [pypy-commit] pypy unicode-dtype: some progress
rlamy
- [pypy-commit] pypy unicode-dtype: hg merge default
rlamy
- [pypy-commit] pypy unicode-dtype: Add comparison operations for unicode
rlamy
- [pypy-commit] pypy unicode-dtype: update test, since unicode is implemented but sorting still isn't
rlamy
- [pypy-commit] pypy unicode-dtype: fix unicode array creation
rlamy
- [pypy-commit] pypy unicode-dtype: Implement UnicodeType.fill()
rlamy
- [pypy-commit] pypy unicode-dtype: Handle str() and repr() on unicode scalars
rlamy
- [pypy-commit] pypy unicode-dtype: Fix repr() of str and unicode scalars
rlamy
- [pypy-commit] pypy unicode-dtype: Add missing UnicodeType methods
rlamy
- [pypy-commit] pypy unicode-dtype: fix translation
rlamy
- [pypy-commit] pypy unicode-dtype: hg merge default
rlamy
- [pypy-commit] pypy numpy-docstrings: Start implementing np.add_docstring
rlamy
- [pypy-commit] pypy numpy-docstrings: Allow adding a docstring to builtin types
rlamy
- [pypy-commit] pypy numpy-docstrings: Handle methods in add_docstring()
rlamy
- [pypy-commit] pypy numpy-docstrings: Handle properties in add_docstring()
rlamy
- [pypy-commit] pypy numpy-docstrings: Create set_docstring() helper, move add_docstring() implementation to app-level and clean up
rlamy
- [pypy-commit] pypy default: vmprof uses x86_64 assembler, so it's only available on x86_64
stefanor
- [pypy-commit] cffi default: Correctly locate py2.7 debug build extensions
stefanor
- [pypy-commit] pypy default: Backed out changeset 5cf9f578ca18
vext01
- [pypy-commit] pypy py3k: fix strerror encoding
yuyichao
Last message date:
Tue Jun 30 19:00:13 CEST 2015
Archived on: Tue Jun 30 19:01:18 CEST 2015
This archive was generated by
Pipermail 0.09 (Mailman edition).