May 2015 Archives by thread
Starting: Fri May 1 07:38:34 CEST 2015
Ending: Sun May 31 22:30:44 CEST 2015
Messages: 1052
- [pypy-commit] pypy default: fix translation
mattip
- [pypy-commit] cffi cffi-1.0: Prepare for the PyPy version: we want to make a similar C extension module,
arigo
- [pypy-commit] jitviewer hierarchy: A quick hack to show the traces in a hierarchical tree.
vext01
- [pypy-commit] cffi cffi-1.0: Add "static" and a test that no unexpected symbols are exported
arigo
- [pypy-commit] jitviewer ui_improvements: Various UI improvements.
vext01
- [pypy-commit] pypy default: or_ to int_or_ in immutable_unique_id
stian
- [pypy-commit] pypy jit_hint_docs: Add more detail to @jit.elidable and @jit.promote.
vext01
- [pypy-commit] pypy jit_hint_docs: Close branch jit_hint_docs
fijal
- [pypy-commit] pypy default: Merged in jit_hint_docs (pull request #322)
fijal
- [pypy-commit] pypy default: Add a comment about @jit.promote_string.
vext01
- [pypy-commit] pypy default: Minor tweak to my recent @jit docstrings.
vext01
- [pypy-commit] pypy default: Fix a couple of tests on Windows
arigo
- [pypy-commit] pypy default: Fix Windows issue left behind by the vmprof branch
arigo
- [pypy-commit] pypy default: clean up
arigo
- [pypy-commit] pypy can_cast: Implement min_dtype() on complex types
rlamy
- [pypy-commit] pypy can_cast: Implement min_dtype() for np.bool_
rlamy
- [pypy-commit] pypy can_cast: Create pypy.module.micronumpy.casting
rlamy
- [pypy-commit] pypy default: Use int with rbigint operations. This provide a upto 25% speedup on such operations, and a minor 5% speedup on pidigits.
stian
- [pypy-commit] pypy default: Merge
stian
- [pypy-commit] cffi cffi-1.0: Add ffi.distutils_extension() and adapt a demo to show the
arigo
- [pypy-commit] cffi cffi-1.0: test fixes
arigo
- [pypy-commit] cffi cffi-1.0: Silence warnings
arigo
- [pypy-commit] pypy default: Fix test on Windows
arigo
- [pypy-commit] pypy default: Adapt for f01fd6fb3a45
arigo
- [pypy-commit] pypy cffi-1.0: A branch in which to add support for cffi 1.0
arigo
- [pypy-commit] pypy cffi-1.0: Add some files
arigo
- [pypy-commit] pypy cffi-1.0: enums, structs
arigo
- [pypy-commit] pypy cffi-1.0: Finish to pass these tests
arigo
- [pypy-commit] pypy numpy-fixes: make astype() return contiguous ndarrays
mattip
- [pypy-commit] pypy numpy-fixes: test for 0-dim array
mattip
- [pypy-commit] pypy numpy-fixes: refactor comparison_func -> bool_result, special case logical_and, logical_or
mattip
- [pypy-commit] cffi cffi-1.0: Add a num_types entry to help pypy's implementation
arigo
- [pypy-commit] pypy cffi-1.0: Add a _nowrapper arg
arigo
- [pypy-commit] pypy cffi-1.0: in-progress
arigo
- [pypy-commit] pypy cffi-1.0: Caching of types inside the _cffi_backend
arigo
- [pypy-commit] pypy cffi-1.0: arrays
arigo
- [pypy-commit] pypy cffi-1.0: function types
arigo
- [pypy-commit] cffi cffi-1.0: Check that the base class' instances don't allow random attributes
arigo
- [pypy-commit] pypy cffi-1.0: Fix tests
arigo
- [pypy-commit] pypy cffi-1.0: ffi.new()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.NULL
arigo
- [pypy-commit] pypy cffi-1.0: more cases in ffi_type()
arigo
- [pypy-commit] pypy cffi-1.0: A few more methods
arigo
- [pypy-commit] pypy cffi-1.0: sizeof, alignof
arigo
- [pypy-commit] pypy cffi-1.0: ffi.error
arigo
- [pypy-commit] pypy numpy-fixes: fix translation
mattip
- [pypy-commit] pypy default: document merged branch
mattip
- [pypy-commit] cffi cffi-1.0: Small refactoring to make a new test pass
arigo
- [pypy-commit] pypy cffi-1.0: Updates to cffi's 47ef4ec2a64c. Add ffi.callback.
arigo
- [pypy-commit] pypy cffi-1.0: ffi.getctype
arigo
- [pypy-commit] pypy cffi-1.0: ffi.addressof()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.new_handle(), ffi.from_handle()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.cast()
arigo
- [pypy-commit] pypy numpy-fixes: fix reduce for any, all
mattip
- [pypy-commit] pypy default: fix own tests
mattip
- [pypy-commit] pypy more-rposix: hg merge default
amauryfa
- [pypy-commit] pypy more-rposix: Remove dead code in ll_os.py. Almost empty now!
amauryfa
- [pypy-commit] pypy more-rposix: Move all test_ll_os to test_rposix
amauryfa
- [pypy-commit] pypy more-rposix: Move tests from ll_os_stat to rposix_stat
amauryfa
- [pypy-commit] pypy more-rposix: Move ll_os_environ to rposix_environ
amauryfa
- [pypy-commit] pypy more-rposix: fix moved symbol
amauryfa
- [pypy-commit] pypy more-rposix: ll_os_path is not used anymore
amauryfa
- [pypy-commit] pypy more-rposix: Move module/ll_time.py to rlib/rtime.py
amauryfa
- [pypy-commit] pypy more-rposix: Fixes
amauryfa
- [pypy-commit] pypy more-rposix: Fix test
amauryfa
- [pypy-commit] pypy more-rposix: Fix for win32
amauryfa
- [pypy-commit] cffi cffi-1.0: Fix manual.c
arigo
- [pypy-commit] cffi cffi-1.0: Beta version 1
arigo
- [pypy-commit] cffi cffi-1.0: fix test
arigo
- [pypy-commit] cffi cffi-1.0: really fix the test
arigo
- [pypy-commit] cffi cffi-1.0: Windows
arigo
- [pypy-commit] cffi cffi-1.0: Windows fixes
arigo
- [pypy-commit] pypy more-rposix: Import rposix &co in extfuncregistry.py.
amauryfa
- [pypy-commit] pypy more-rposix: Remove all code related to "lazy registration" of functions, not used anymore.
amauryfa
- [pypy-commit] pypy more-rposix: RPython really replaces time.time by the version in rtime.py:
amauryfa
- [pypy-commit] pypy default: Bumped greenlet version -- the upstream version has no public API changes
alex_gaynor
- [pypy-commit] pypy default: merged upstream
alex_gaynor
- [pypy-commit] pypy more-rposix: Fix failure in test_extfunc
amauryfa
- [pypy-commit] pypy remove-frame-debug-attrs: Initial commit - a branch to try and remove the debug attributes on frames
fijal
- [pypy-commit] pypy default: Fix test when run alone
arigo
- [pypy-commit] pypy remove-frame-debug-attrs: fix remaining cases
fijal
- [pypy-commit] pypy remove-frame-debug-attrs: fixup pypyjit module
fijal
- [pypy-commit] pypy remove-frame-debug-attrs: oops
fijal
- [pypy-commit] pypy remove-frame-debug-attrs: fix rpython
fijal
- [pypy-commit] pypy.org extradoc: update the values
arigo
- [pypy-commit] pypy default: Skip vmprof tests on CPUs different than x86-64
arigo
- [pypy-commit] pypy remove-frame-debug-attrs: a bit blindly fix cpyext
fijal
- [pypy-commit] pypy remove-frame-debug-attrs: close to be merged branch
fijal
- [pypy-commit] pypy default: (fijal, cfbolz) remove some attributes from frames and put them on the
fijal
- [pypy-commit] pypy default: make debugdata a virtualizable field
fijal
- [pypy-commit] pypy default: give up on static linking of libunwind
fijal
- [pypy-commit] pypy default: Issue #1194
arigo
- [pypy-commit] pypy default: move w_locals to the debug subobject
fijal
- [pypy-commit] pypy default: Encode frame_finished_execution in last_instr == -2
fijal
- [pypy-commit] pypy conditional_call_value: work in progress
fijal
- [pypy-commit] pypy conditional_call_value: I don't know what this thing does anyway
fijal
- [pypy-commit] pypy conditional_call_value: This is how you're supposed to use "callback". It's very obscure...
arigo
- [pypy-commit] pypy default: Try to document the complicated API of emulate_pbc_call().
arigo
- [pypy-commit] pypy default: The point of unique_key
arigo
- [pypy-commit] pypy conditional_call_value: A very simple test to add a conditional_call_value
fijal
- [pypy-commit] cffi cffi-1.0: Tweak
arigo
- [pypy-commit] pypy conditional_call_value: support cond_call_value in the x86 backend (I hope)
fijal
- [pypy-commit] pypy conditional_call_value: try to use conditional_call_value - NOTE THAT WE IGNORED ELIDABLE here
fijal
- [pypy-commit] pypy conditional_call_value: mention it in a comment
fijal
- [pypy-commit] pypy default: Support for Windows icons
arigo
- [pypy-commit] pypy default: Add pypy.ico from Eun Che on pypy-dev
arigo
- [pypy-commit] pypy default: merge heads
arigo
- [pypy-commit] pypy conditional_call_value: I have probably no idea what I'm doing, but try to support a few more cases
fijal
- [pypy-commit] pypy conditional_call_value: write a test, does not help
fijal
- [pypy-commit] pypy conditional_call_value: write (crashing) test_ll_random
fijal
- [pypy-commit] pypy conditional_call_value: an obvious fix, test_zll_random stil lfails for a bit unclear reasons
fijal
- [pypy-commit] pypy default: "import posix" fails actually on Windows.
arigo
- [pypy-commit] pypy default: darwin fix
arigo
- [pypy-commit] cffi default: Silence msvc warnings
arigo
- [pypy-commit] pypy vecopt2: new vecopt branch
plan_rich
- [pypy-commit] pypy vecopt2: added a file to implement the dependency graph and the first failing test for it
plan_rich
- [pypy-commit] pypy vecopt2: added simple version that calcualtes the dependencies on a ssa trace. added test to check if label definition creates dependencies correctly
plan_rich
- [pypy-commit] pypy vecopt2: started to unroll a trace in the optimizer. work in progress
plan_rich
- [pypy-commit] pypy vecopt2: passing first unroll test
plan_rich
- [pypy-commit] pypy vecopt2: reorderd some methods in unroll
plan_rich
- [pypy-commit] pypy vecopt2: added information on how the optimzer internally works
plan_rich
- [pypy-commit] pypy vecopt2: indention for syntax highlightning
plan_rich
- [pypy-commit] pypy vecopt2: moved around my optimizer files, unrolling does not get it' own file
plan_rich
- [pypy-commit] pypy vecopt2: forgot to attach the new files, added another unrolling test case and added some comments to the test cases
plan_rich
- [pypy-commit] pypy vecopt2: making the backend ready to be translated. stumbled over several problems not allowed in rpython (slice[1:-1], returning multiple values)
plan_rich
- [pypy-commit] pypy vecopt2: pair constructor (testing hg branch)
plan_rich
- [pypy-commit] pypy vecopt2: creating basic classes needed
plan_rich
- [pypy-commit] pypy vecopt2: zip is not available in rpython
plan_rich
- [pypy-commit] pypy vecopt2: forgot range(
plan_rich
- [pypy-commit] pypy vecopt2: working on packing instructions
plan_rich
- [pypy-commit] pypy vecopt2: added test cases for the dependency graph and testing the calc of adjacent memory locations (not working yet)
plan_rich
- [pypy-commit] pypy vecopt2: impl & test memory adjacent calculation
plan_rich
- [pypy-commit] pypy vecopt2: added some more test cases. int_sub/int_mul is not supported to calculate the memory ref positions
plan_rich
- [pypy-commit] pypy vecopt2: added missing vectorize flag for test cases. all passed in jit/metainterp
plan_rich
- [pypy-commit] pypy vecopt2: extended memory ref analysis to consider division as well, added some tests
plan_rich
- [pypy-commit] pypy vecopt2: adjusted the test suit after a refactoring the vectorization optimizer
plan_rich
- [pypy-commit] pypy vecopt2: debug_merge_points are not unrolled, but only prepended & appended
plan_rich
- [pypy-commit] pypy vecopt2: separated dependency graph testing from the vectorize optimization tests.
plan_rich
- [pypy-commit] pypy vecopt2: enhanced the vectorizing testcase, clarified unroll count. it is now a number how often to fruther unroll it, not the total amount
plan_rich
- [pypy-commit] pypy vecopt2: removed readme in optimizeopt -> was merged in doc
plan_rich
- [pypy-commit] pypy vecopt2: unrolling now keeps track of the fail arguments and renames them correctly
plan_rich
- [pypy-commit] pypy vecopt2: added new dependency changes
plan_rich
- [pypy-commit] pypy vecopt2: extended the instruction packing. it now finds adjacent memory references and packs them into pairs in the packset
plan_rich
- [pypy-commit] pypy vecopt2: extended the test cases and removed a undiscovered bug in the unrolling/renaming phase
plan_rich
- [pypy-commit] pypy vecopt2: added failing test case for extending pack sets
plan_rich
- [pypy-commit] pypy vecopt2: impl. follow def use chain. packset is extended by independent follow up instructions that reuse the definition
plan_rich
- [pypy-commit] pypy vecopt2: impl. follow use def chains (similar to def use chains)
plan_rich
- [pypy-commit] pypy vecopt2: making the dependency builder less conservative. added test for aliasing modification problem
plan_rich
- [pypy-commit] pypy vecopt2: added a test for an unalias counter example
plan_rich
- [pypy-commit] pypy vecopt2: moved integral modification and memory ref to dependency.py. this is needed to get rid of dependencies between 2 set array at different indices (work in progress)
plan_rich
- [pypy-commit] pypy vecopt2: dependency graph now tracks array modifications and discards edges of proven cell access to not overlap
plan_rich
- [pypy-commit] pypy vecopt2: packset combination (need to rewrite for rpython) and tests
plan_rich
- [pypy-commit] pypy vecopt2: added test with more packs
plan_rich
- [pypy-commit] pypy vecopt2: added a scheduling test (failing)
plan_rich
- [pypy-commit] pypy vecopt2: added vector IR operations (load/store & arithmetic)
plan_rich
- [pypy-commit] pypy vecopt2: added vector integer field to resop class. removed some specific vec ops (descr might be used for that)
plan_rich
- [pypy-commit] pypy vecopt2: enhanced dependency test. no boiler plate code to define dependencies (but annotate in the code instead)
plan_rich
- [pypy-commit] pypy vecopt2: all tests passing after finishing the refactoring
plan_rich
- [pypy-commit] pypy vecopt2: introducing box_vector as new boxes (work in progress)
plan_rich
- [pypy-commit] pypy vecopt2: BoxVector is now used as vector variable
plan_rich
- [pypy-commit] pypy vecopt2: memory reference now correctly tracks if bytes overlap (previously the cell stride was ignored)
plan_rich
- [pypy-commit] pypy vecopt2: impl. llgraph vector instructions dispatch
plan_rich
- [pypy-commit] pypy vecopt2: starting to modify the dependency construction to track all integral modifications
plan_rich
- [pypy-commit] pypy vecopt2: updated tests to ignore non present transitive dependencies, nearly completed the new integral forward modification migration
plan_rich
- [pypy-commit] pypy vecopt2: finished index variable tracking, added a new guard (GUARD_NO_EARLY_EXIT)
plan_rich
- [pypy-commit] pypy vecopt2: work in progress refactoring dependencies to easier remove instructions
plan_rich
- [pypy-commit] pypy vecopt2: work in progress (continued)
plan_rich
- [pypy-commit] pypy vecopt2: the dependency graph now wraps each operation in a Node object. This makes the arch. much cleaner and separates concerns
plan_rich
- [pypy-commit] pypy vecopt2: updated the vectorizer to use the new dependency graph, not yet finished (simplifications included)
plan_rich
- [pypy-commit] pypy vecopt2: relaxing guards dependency works for the first simple case
plan_rich
- [pypy-commit] pypy vecopt2: extend packset now checks both sides of the pack
plan_rich
- [pypy-commit] pypy vecopt2: tested and adapted the trace optimizer for loops.
plan_rich
- [pypy-commit] pypy vecopt2: vec operations always include the amount of items to process, adapted tests
plan_rich
- [pypy-commit] pypy vecopt2: introduced vec_getarrayitem_raw which was previously handled by vector raw load (was not correct)
plan_rich
- [pypy-commit] pypy vecopt2: removed print statements for a translated binary (tests still print some verbose informations)
plan_rich
- [pypy-commit] pypy vecopt2: renamed VectorizeTest class to plural
plan_rich
- [pypy-commit] pypy vecopt2: renamed guard_no_early_exit to guard_early_exit
plan_rich
- [pypy-commit] pypy vecopt2: x86_64 backend is now capable of emitting sse2 instructions for the current vector operations
plan_rich
- [pypy-commit] pypy vecopt2: added vec_int_signext to the backend (is ignored and register is forced)
plan_rich
- [pypy-commit] pypy vecopt2: wrapping all parameters passed to the assembler (from regalloc) in imm
plan_rich
- [pypy-commit] pypy vecopt2: changes to make the rtyper work correctly, SIMD loads now only from is now aligned (not correct, just for testing)
plan_rich
- [pypy-commit] pypy vecopt2: started to implement guard folding. (j=i+1, guard(j), k=j+1, guard(k) => j=i+2, guard(j))
plan_rich
- [pypy-commit] pypy vecopt2: vectorization now uses the preamble of the unrolling optimization, this is a step towards a unified unrolling algorithm (and keeps most of the variables in the register)
plan_rich
- [pypy-commit] pypy vecopt2: added a new test to collapse guards, I plan to restructure this and make it on the level of a dependency graph
plan_rich
- [pypy-commit] pypy vecopt2: weaker guards are stripped from the trace
plan_rich
- [pypy-commit] pypy vecopt2: guard_early_exit are not passed to the backend. vectorize tests all pass again
plan_rich
- [pypy-commit] pypy vecopt2: changes to make rpython happy
plan_rich
- [pypy-commit] pypy vecopt2: bridge can now be assembled (previously an early exit skipped to the blackhole interpreter)
plan_rich
- [pypy-commit] pypy vecopt2: updated vector to small test (simpler) and added new test for constant expansion
plan_rich
- [pypy-commit] pypy vecopt2: removed check trace count == 1, added missed changes in merge
plan_rich
- [pypy-commit] pypy vecopt2: turned off vectorization by default
plan_rich
- [pypy-commit] pypy vecopt2: removed vectorized parameter of jit driver (micronumpy)
plan_rich
- [pypy-commit] pypy vecopt2: merged default
plan_rich
- [pypy-commit] pypy vecopt2: assert isinstance of none prevented (for tests)
plan_rich
- [pypy-commit] pypy vecopt: renamed branch to vecopt
plan_rich
- [pypy-commit] pypy vecopt: activated vectorize for 2 jit drivers (micronumpy)
plan_rich
- [pypy-commit] pypy conditional_call_value: consider COND_CALL_VALUE in heapcache
fijal
- [pypy-commit] pypy default: port the cpython test and improve situation a little bit
fijal
- [pypy-commit] pypy default: Issue 1194: test and fix
arigo
- [pypy-commit] pypy default: a slightly different fix that actually fixes the problem
fijal
- [pypy-commit] pypy vecopt: test commit, have i resolved all my hg problems?
plan_rich
- [pypy-commit] pypy default: Issue #2039
arigo
- [pypy-commit] cffi cffi-1.0: Make 'build' the default
arigo
- [pypy-commit] pypy conditional_call_value: write some tests
fijal
- [pypy-commit] pypy default: sounds like a sane assert
fijal
- [pypy-commit] pypy numpy-fixes: add __array_priority__ which should determine the return type for ufuncs
mattip
- [pypy-commit] pypy numpy-fixes: test, fix obscure use of __array_priority__ in classes like numpy.Polynomial
mattip
- [pypy-commit] pypy default: backout e24b51be112d to try and resolve the problem
fijal
- [pypy-commit] pypy default: fix the backout
fijal
- [pypy-commit] cffi cffi-1.0: Revert this version number change, which doesn't make sense
arigo
- [pypy-commit] pypy can_cast: hg merge default
rlamy
- [pypy-commit] pypy can_cast: document some equivalences with numpy's C source
rlamy
- [pypy-commit] pypy numpy-fixes: merge default into branch
mattip
- [pypy-commit] pypy vecopt: updated fake objects to be able to run micro numpy code (not finished yet)
plan_rich
- [pypy-commit] pypy vecopt: test_zjit up and running again (extended the fake space)
plan_rich
- [pypy-commit] pypy conditional_call_value: fix the test
fijal
- [pypy-commit] cffi default: Fix
arigo
- [pypy-commit] cffi cffi-1.0: hg merge default
arigo
- [pypy-commit] pypy conditional_call_value: try to write elidable COND_CALL, will write tests
fijal
- [pypy-commit] pypy conditional_call_value: pfff rpython
fijal
- [pypy-commit] pypy can_cast: Add missing comparison operations on dtype
rlamy
- [pypy-commit] pypy default: Add a passing test, with a theory about how it could future-proof
arigo
- [pypy-commit] pypy default: Failing test
arigo
- [pypy-commit] pypy default: Fix for 078cf2a4aa22
arigo
- [pypy-commit] pypy can_cast: Fix typeinfo dict
rlamy
- [pypy-commit] pypy numpy-fixes: move test and make 32-bit friendly, revert string comparison on cmp-func execption
mattip
- [pypy-commit] pypy numpy-fixes: ajdust expected jit ops for changes in loop.py, ufunc.py
mattip
- [pypy-commit] pypy default: Test and fix for gdbm
arigo
- [pypy-commit] pypy default: The gdbm library is not thread-safe. Add a global lock.
arigo
- [pypy-commit] pypy default: don't create debug object just because we used a local import
fijal
- [pypy-commit] cffi cffi-1.0: To-do
arigo
- [pypy-commit] cffi cffi-1.0: Bump the version to 1.0.0b2
arigo
- [pypy-commit] pypy vecopt: rewritten parts of scheduling (priority can be set to nodes)
plan_rich
- [pypy-commit] pypy vecopt: updated tests (some of them needed adjusting after scheduling changed)
plan_rich
- [pypy-commit] pypy vecopt: numpy call2 now runs as vector ops (llgraph, x86 not yet)
plan_rich
- [pypy-commit] cffi cffi-1.0: Increase laziness in a small case.
arigo
- [pypy-commit] pypy cffi-1.0: Import the test_c.py file (it wasn't really changed)
arigo
- [pypy-commit] pypy cffi-1.0: jit tweak
arigo
- [pypy-commit] pypy cffi-1.0: Finish to port the changes to c/_cffi_backend
arigo
- [pypy-commit] pypy cffi-1.0: Tweaks to end up in _cffi_backend when loading C extension modules
arigo
- [pypy-commit] pypy cffi-1.0: fight fight
arigo
- [pypy-commit] pypy default: a slightly different fix
fijal
- [pypy-commit] pypy conditional_call_value: finish a test and fix
fijal
- [pypy-commit] pypy can_cast: Add np.min_scalar_type()
rlamy
- [pypy-commit] pypy cffi-1.0: in-progress: adding recompiler tests in pypy
arigo
- [pypy-commit] cffi cffi-1.0: PyPy tweak
arigo
- [pypy-commit] cffi cffi-1.0: ffi.emit_c_code()
arigo
- [pypy-commit] cffi cffi-1.0: Generate pypy-friendly code, following the example of manual.c
arigo
- [pypy-commit] pypy cffi-1.0: Starting on lib_obj
arigo
- [pypy-commit] pypy cffi-1.0: Tweaks
arigo
- [pypy-commit] pypy cffi-1.0: Now the tests start passing
arigo
- [pypy-commit] pypy default: oups in three files' typedef's acceptable_as_base_class
arigo
- [pypy-commit] pypy default: Replace some "acceptable_as_base_class=False" with "assert not
arigo
- [pypy-commit] cffi cffi-1.0: simplify
arigo
- [pypy-commit] cffi cffi-1.0: Add another future proofing
arigo
- [pypy-commit] cffi cffi-1.0: Move test
arigo
- [pypy-commit] pypy cffi-1.0: Globals
arigo
- [pypy-commit] pypy cffi-1.0: Typenames
arigo
- [pypy-commit] pypy cffi-1.0: fix more tests
arigo
- [pypy-commit] pypy cffi-1.0: Writing to global vars
arigo
- [pypy-commit] pypy cffi-1.0: Needs to import _cffi_backend at app-level, otherwise the startup()
arigo
- [pypy-commit] pypy cffi-1.0: error cases of realize_global_int
arigo
- [pypy-commit] pypy cffi-1.0: non-integer constants
arigo
- [pypy-commit] pypy cffi-1.0: repr, dir on Lib
arigo
- [pypy-commit] cffi cffi-1.0: Test repr(lib)
arigo
- [pypy-commit] cffi cffi-1.0: Use a '_CFFI_' prefix in test names to avoid clashes
arigo
- [pypy-commit] pypy cffi-1.0: new feature in TypeDefs: you can use ClassAttr(f) to make a class
arigo
- [pypy-commit] pypy cffi-1.0: use ClassAttr
arigo
- [pypy-commit] pypy vecopt: renamed zjit_test,
plan_rich
- [pypy-commit] pypy vecopt: updated test for constant expansion (test_zjit),
plan_rich
- [pypy-commit] pypy vecopt: unpack/pack operations (vector[x]->scalar, scalar->vector[y])
plan_rich
- [pypy-commit] cffi cffi-1.0: Add keyword arguments to a few ffi methods
arigo
- [pypy-commit] pypy cffi-1.0: fix test
arigo
- [pypy-commit] pypy cffi-1.0: More test fixing
arigo
- [pypy-commit] cffi cffi-1.0: Test for f9eca71a085b
arigo
- [pypy-commit] pypy cffi-1.0: opaque structs
arigo
- [pypy-commit] pypy cffi-1.0: ffi.offsetof(), and improve JITting of ffi.addressof()
arigo
- [pypy-commit] pypy cffi-1.0: more progress
arigo
- [pypy-commit] pypy cffi-1.0: Grumble, figured out that the reason we get sometimes unexpected objects
arigo
- [pypy-commit] pypy cffi-1.0: Enums
arigo
- [pypy-commit] pypy cffi-1.0: Pff, this fixes a leak
arigo
- [pypy-commit] pypy cffi-1.0: unnamed structs
arigo
- [pypy-commit] pypy cffi-1.0: fix test_module_name_in_package
arigo
- [pypy-commit] pypy numpy-fixes: mark places where __array_wrap__ should be called
mattip
- [pypy-commit] pypy numpy-fixes: add failing test, support call2; in ufuncs call __array_wrap__ after __array_finalize__
mattip
- [pypy-commit] pypy numpy-fixes: fix for flatiter
mattip
- [pypy-commit] pypy numpy-fixes: fix the hack to fix tests
mattip
- [pypy-commit] pypy cffi-1.0: Prepare tests for ffi.include()
arigo
- [pypy-commit] pypy cffi-1.0: next include test passing
arigo
- [pypy-commit] pypy cffi-1.0: Next test
arigo
- [pypy-commit] pypy cffi-1.0: Remaining include tests
arigo
- [pypy-commit] pypy cells-local-stack: merge default
cfbolz
- [pypy-commit] cffi cffi-1.0: comment
arigo
- [pypy-commit] cffi cffi-1.0: Add a passing test
arigo
- [pypy-commit] pypy cffi-1.0: Test and fix
arigo
- [pypy-commit] pypy cffi-1.0: Translation fixes (yay test_ztranslation)
arigo
- [pypy-commit] pypy cells-local-stack: cells field is no longer needed
cfbolz
- [pypy-commit] pypy cells-local-stack: fix asserts about the stack depth
cfbolz
- [pypy-commit] pypy cells-local-stack: remove unused methods
cfbolz
- [pypy-commit] pypy cells-local-stack: fix virtualizable declaration
cfbolz
- [pypy-commit] pypy cffi-1.0: JIT fixes
arigo
- [pypy-commit] pypy cffi-1.0: NonConstant is not actually needed here (rffi.cast() is enough) and
arigo
- [pypy-commit] cffi cffi-1.0: 'cffi_modules' can be unicodes on 2.7 with unicode_literals
arigo
- [pypy-commit] cffi cffi-1.0: unicode_literals detection
arigo
- [pypy-commit] cffi cffi-1.0: Improve the test for unicode_literals. Now it fails
arigo
- [pypy-commit] cffi cffi-1.0: Accept unicode literals for type specifications too
arigo
- [pypy-commit] cffi cffi-1.0: add a doc to-do
arigo
- [pypy-commit] pypy cffi-1.0: @elidable needs a bit of care: it cannot be used on functions that
arigo
- [pypy-commit] pypy cffi-1.0: Accept unicode literals in type declarations
arigo
- [pypy-commit] pypy fix-result-types: move some stuff
rlamy
- [pypy-commit] pypy cffi-1.0: More dont_look_inside to avoid confusing the JIT
arigo
- [pypy-commit] pypy cffi-1.0: The struct names should not include the "struct " part here.
arigo
- [pypy-commit] pypy default: Issue #2043 fix
arigo
- [pypy-commit] pypy cells-local-stack: fix translation
cfbolz
- [pypy-commit] pypy cffi-1.0: Import cffi 1.0.0 (revision 8bb4a702b17d)
arigo
- [pypy-commit] pypy cffi-1.0: Update cffi.egg-info to the 1.1 version, which allows "entry_points.txt"
arigo
- [pypy-commit] cffi cffi-1.0: Add TODO
arigo
- [pypy-commit] cffi cffi-1.0: Python 3.2 support
arigo
- [pypy-commit] cffi cffi-1.0: Skip this test on pypy
arigo
- [pypy-commit] cffi cffi-1.0: merge heads
arigo
- [pypy-commit] cffi cffi-1.0: Passing test
arigo
- [pypy-commit] cffi cffi-1.0: Failed at manually sorting ;-)
arigo
- [pypy-commit] cffi cffi-1.0: A few more small tests
arigo
- [pypy-commit] pypy cffi-1.0: Test for keyword arguments
arigo
- [pypy-commit] pypy cffi-1.0: OP_BITFIELD
arigo
- [pypy-commit] pypy cffi-1.0: ffi.buffer()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.from_buffer()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.CData, ffi.CType
arigo
- [pypy-commit] pypy cffi-1.0: ffi.gc()
arigo
- [pypy-commit] pypy cffi-1.0: ffi.getwinerror() (untested yet)
arigo
- [pypy-commit] cffi cffi-1.0: The ``cffi_modules = ["module:ffi"]`` can now also name a global
arigo
- [pypy-commit] cffi cffi-1.0: Test and fix
arigo
- [pypy-commit] cffi cffi-1.0: Typo (thanks lazka)
arigo
- [pypy-commit] pypy fix-result-types: Start reimplementing np.result_type() to make it compatible with cnumpy
rlamy
- [pypy-commit] pypy can_cast: Close branch can_cast
rlamy
- [pypy-commit] pypy default: Merged in can_cast (pull request #324)
rlamy
- [pypy-commit] pypy fix-result-types: Use find_result_type() instead of find_binop_result_dtype() in a few places
rlamy
- [pypy-commit] cffi cffi-1.0: improve the test
arigo
- [pypy-commit] cffi cffi-1.0: one more test
arigo
- [pypy-commit] cffi cffi-1.0: Return structs via a hidden pointer argument
arigo
- [pypy-commit] cffi cffi-1.0: Fix tests
arigo
- [pypy-commit] pypy fix-result-types: hg merge default
rlamy
- [pypy-commit] pypy cffi-1.0: Test and fix fix fix in progress
arigo
- [pypy-commit] pypy cffi-1.0: Struct/union returns
arigo
- [pypy-commit] pypy cffi-1.0: import cffi/c/test_c.py
arigo
- [pypy-commit] pypy cffi-1.0: Fixes
arigo
- [pypy-commit] pypy cffi-1.0: Forgot to add '...'
arigo
- [pypy-commit] pypy cffi-1.0: @edliable fix: can't call _new_function_type() from an elidable function
arigo
- [pypy-commit] cffi cffi-1.0: Improve the error message when dependent modules cannot be loaded
arigo
- [pypy-commit] pypy cffi-1.0: Improve the error message when dependent modules cannot be loaded
arigo
- [pypy-commit] cffi cffi-1.0: add test
arigo
- [pypy-commit] pypy numpy-fixes: cleanup and call __array_wrap__ where needed, also add missing subtype compatability on reduce()
mattip
- [pypy-commit] pypy numpy-fixes: fix, own tests now pass
mattip
- [pypy-commit] pypy cffi-1.0: A large amount of efforts in a refactoring for two extra lines of test
arigo
- [pypy-commit] pypy cffi-1.0: Translation fix
arigo
- [pypy-commit] pypy cffi-1.0: fix
arigo
- [pypy-commit] cffi cffi-1.0: change the pypy exported version number
arigo
- [pypy-commit] pypy cffi-1.0: Change the versions to include "2.6" in it, which is the next pypy version;
arigo
- [pypy-commit] pypy cffi-1.0: merge heads
arigo
- [pypy-commit] cffi cffi-1.0: Rename the hidden argument 'result'
arigo
- [pypy-commit] cffi default: Support doing a bit more things with a "typedef struct { } *name; "
arigo
- [pypy-commit] cffi cffi-1.0: Test and fix: don't write <ctype '1 *'> but <ctype 'struct $1 *'>
arigo
- [pypy-commit] cffi cffi-1.0: Hack hack hack to support test_include_5
arigo
- [pypy-commit] cffi cffi-1.0: Fix issues with uncomputed alignment
arigo
- [pypy-commit] pypy cffi-1.0: hg merge default
arigo
- [pypy-commit] pypy cffi-1.0: Trying it out
arigo
- [pypy-commit] pypy cffi-1.0: Update to cffi/d09c84106ff9
arigo
- [pypy-commit] pypy cffi-1.0: Windows compat
arigo
- [pypy-commit] cffi cffi-1.0: Distutils fun on Windows
arigo
- [pypy-commit] pypy cffi-1.0: Don't hard-code gcc here
arigo
- [pypy-commit] pypy cffi-1.0: Tweak tweak
arigo
- [pypy-commit] pypy cffi-1.0: Next fix
arigo
- [pypy-commit] pypy default: Write a small comment on the same line that points to the long
arigo
- [pypy-commit] pypy fix-result-types: begin chopping up find_binop_result_dtype()
rlamy
- [pypy-commit] pypy default: Instead of crashing when we don't find _reuse/_drop, display a warning.
arigo
- [pypy-commit] pypy numpy-fixes: test, fix for arange(array([10]))
mattip
- [pypy-commit] pypy numpy-fixes: adjust jit ops (again)
mattip
- [pypy-commit] pypy default: fix test
mattip
- [pypy-commit] pypy fix-result-types: small cleanup
rlamy
- [pypy-commit] pypy fix-result-types: kill _get_dtype()
rlamy
- [pypy-commit] pypy fix-result-types: extract method W_Ufunc{1, 2}.call_scalar()
rlamy
- [pypy-commit] cffi cffi-1.0: Partial test and fix, explaining the limitation
arigo
- [pypy-commit] pypy cffi-1.0: Partial test and fix, explaining the limitation
arigo
- [pypy-commit] cffi cffi-1.0: Start reviewing the docs
arigo
- [pypy-commit] cffi cffi-1.0: Tweak tweak tweak: use an initialization sequence on CPython that is closer
arigo
- [pypy-commit] pypy cffi-1.0: Passing test
arigo
- [pypy-commit] cffi cffi-1.0: Python 3 compat
arigo
- [pypy-commit] cffi cffi-1.0: A case where raising NotImplementedError is more appropriate than just
arigo
- [pypy-commit] cffi cffi-1.0: Make ffi->types_builder an inlined struct; add a ref from the lib to the ffi
arigo
- [pypy-commit] cffi cffi-1.0: Reasonably messy to implement: ffi.addressof(lib, "var")
arigo
- [pypy-commit] pypy vecopt: renamed detect_sse2.py to detect_feature.py
plan_rich
- [pypy-commit] pypy vecopt: simplifications, added a x86_64 sse4 cpu
plan_rich
- [pypy-commit] pypy vecopt: added some missing vector x86 instructions to mc
plan_rich
- [pypy-commit] pypy cffi-1.0: ffi.addressof(lib, "var")
arigo
- [pypy-commit] cffi cffi-1.0: in-progress: ffi.dlopen()
arigo
- [pypy-commit] cffi cffi-1.0: in-progress
arigo
- [pypy-commit] cffi cffi-1.0: Just enough to pass this small demo
arigo
- [pypy-commit] cffi cffi-1.0: struct/unions
arigo
- [pypy-commit] cffi cffi-1.0: enums, integer constants
arigo
- [pypy-commit] cffi cffi-1.0: typenames
arigo
- [pypy-commit] pypy fix-result-types: extract W_Ufunc1.find_specialization()
rlamy
- [pypy-commit] pypy fix-result-types: Fix casting table
rlamy
- [pypy-commit] pypy fix-result-types: Use the same logic as cnumpy in W_Ufunc1.find_specialization()
rlamy
- [pypy-commit] cffi cffi-1.0: Fix
arigo
- [pypy-commit] cffi cffi-1.0: Hum, we need to close manually the subprocess.PIPE it seems
arigo
- [pypy-commit] cffi cffi-1.0: Fix for 3.4.1 (this version forces -Werror=declaration-after-statement)
arigo
- [pypy-commit] cffi cffi-1.0: Move the remaining interesting parts of the PLAN file to doc/
arigo
- [pypy-commit] cffi cffi-1.0: Move the grant proposal to doc/
arigo
- [pypy-commit] cffi cffi-1.0: Move into a misc/ subdir
arigo
- [pypy-commit] cffi cffi-1.0: the big Moving Files Around step
arigo
- [pypy-commit] cffi cffi-1.0: more fixes
arigo
- [pypy-commit] cffi cffi-1.0: Update MANIFEST.in
arigo
- [pypy-commit] pypy.org extradoc: Update the module lists
arigo
- [pypy-commit] cffi cffi-1.0: Getting started on 'recompiler' producing a python file for dlopen()
arigo
- [pypy-commit] cffi cffi-1.0: Make "static const int FOO = VALUE; " fully equivalent to "#define FOO VALUE"
arigo
- [pypy-commit] pypy vecopt: vector add now passing for double floating points
plan_rich
- [pypy-commit] pypy vecopt: added integer/float types to zjit test (int 8, 16, 32, 64, float 32, 64)
plan_rich
- [pypy-commit] cffi cffi-1.0: Next test passes
arigo
- [pypy-commit] cffi cffi-1.0: enum
arigo
- [pypy-commit] pypy default: when escaping an array, remember its length
cfbolz
- [pypy-commit] pypy numpy-fixes: test, fix array bound overflow caused by not recalculating backstrides
mattip
- [pypy-commit] pypy default: test, fix for issue #2046
mattip
- [pypy-commit] pypy numpy-fixes: document branch
mattip
- [pypy-commit] pypy cells-local-stack: close to-be-merged branch
cfbolz
- [pypy-commit] pypy default: merge cells-local-stack
cfbolz
- [pypy-commit] pypy default: merge default
cfbolz
- [pypy-commit] pypy vecopt: added many opcodes for x86 that operate on packed data (single float)
plan_rich
- [pypy-commit] pypy default: merge numpy-fixes which further advances numpy complience
mattip
- [pypy-commit] cffi cffi-1.0: Windows fix
arigo
- [pypy-commit] cffi cffi-1.0.0b3: Fix for 3.4.1 (this version forces -Werror=declaration-after-statement)
arigo
- [pypy-commit] pypy fix-result-types: cleanup
rlamy
- [pypy-commit] pypy fix-result-types: move more stuff inside find_specialization()
rlamy
- [pypy-commit] pypy fix-result-types: first step towards computing the loop's output type in _calc_dtype()
rlamy
- [pypy-commit] pypy fix-result-types: support 'casting' argument in unary ufuncs
rlamy
- [pypy-commit] pypy win32-optionals: start branch to support pythonw.exe, largeaddressaware on win32
mattip
- [pypy-commit] pypy win32-optionals: use sys.platform instead of 'host' in --platform to utilize Option.requires
mattip
- [pypy-commit] pypy win32-optionals: improve error message if option violates require= kwarg, add largeaddressaware, pass tests
mattip
- [pypy-commit] pypy default: Updated the embedding documentation to contain only the modern way we
arigo
- [pypy-commit] pypy vecopt: float32/64 addition of vectors now uses packed vector load x86
plan_rich
- [pypy-commit] pypy default: test, fix? nditer for negative strides
mattip
- [pypy-commit] pypy pythonoptimize-env: test, fix use of PYTHONOPTIMIZE in env, issue #2044
mattip
- [pypy-commit] pypy pythonoptimize-env: cleanup (cfbolz)
mattip
- [pypy-commit] pypy default: reduce warnings, add labels to branches
mattip
- [pypy-commit] pypy default: fix test, specify int size when examining strides
mattip
- [pypy-commit] pypy default: Mention these two things here (from pypy.org/source/compat.rst)
arigo
- [pypy-commit] pypy.org extradoc: Rewrite a bit this section to focus on the commonly-encountered issue
arigo
- [pypy-commit] pypy default: Issue #2048
arigo
- [pypy-commit] pypy default: This loop was infinite if it reaches '--end--'
arigo
- [pypy-commit] pypy vecopt: resolved an issue that would generate wrong packing immediates for shufps.
plan_rich
- [pypy-commit] pypy default: Kill 'exchange_result_libffi' and tweak jit_libffi. This might actually
arigo
- [pypy-commit] pypy default: Add an underscore in front of functions that are not meant to be exposed
arigo
- [pypy-commit] pypy default: Tweak: on 64 bits 3/4rd of the tests were skipped
arigo
- [pypy-commit] pypy default: more tests
arigo
- [pypy-commit] pypy numpy-fixes: test, fix creating ndarray from buffer, which does [ord(c) for c in buffer]
mattip
- [pypy-commit] pypy pythonoptimize-env: merge default into branch
mattip
- [pypy-commit] pypy fix-result-types: simplify loop.call1()
rlamy
- [pypy-commit] pypy fix-result-types: simplify loop.call2()
rlamy
- [pypy-commit] pypy.org extradoc: Remove "sandboxing" from the front page of pypy.org. It's still
arigo
- [pypy-commit] pypy numpy-fixes: a.view() should respect the subtype of a
mattip
- [pypy-commit] pypy pythonoptimize-env: missed import
mattip
- [pypy-commit] pypy vecopt: split box_(un)pack into float_(un)pack and int_(un)pack, adjusted some tests that unrolled to often (scheduler splits packed instructions if size is too big)
plan_rich
- [pypy-commit] pypy vecopt: typed expand to float/int
plan_rich
- [pypy-commit] pypy fix-result-types: fix return type for unary ufuncs on object scalars
rlamy
- [pypy-commit] pypy default: Two new performance tests, with corresponding fixes (likely broken by 749bf9a13d9c)
arigo
- [pypy-commit] cffi cffi-1.0: Structs
arigo
- [pypy-commit] cffi cffi-1.0: fixes
arigo
- [pypy-commit] cffi cffi-1.0: Various tweaks until we get some more-or-less complete dlopen() support
arigo
- [pypy-commit] cffi cffi-1.0: more tests
arigo
- [pypy-commit] cffi cffi-1.0: Includes of dlopen-style ffis. Negative constants in "static const int = ...".
arigo
- [pypy-commit] cffi cffi-1.0: Test and fix for struct names from included ffis
arigo
- [pypy-commit] cffi cffi-1.0: Clean up
arigo
- [pypy-commit] cffi cffi-1.0: Add a test
arigo
- [pypy-commit] cffi cffi-1.0: Array lengths
arigo
- [pypy-commit] cffi cffi-1.0: Modernize this dlopen() demo
arigo
- [pypy-commit] cffi cffi-1.0: extra test
arigo
- [pypy-commit] cffi cffi-1.0: Bump the internal version number
arigo
- [pypy-commit] cffi cffi-1.0: ffi.dlclose(). Global variables.
arigo
- [pypy-commit] cffi cffi-1.0: Windows compat (py2 only)
arigo
- [pypy-commit] pypy fix-result-types: Use the call_scalar() fast path only when a scalar is passed in
rlamy
- [pypy-commit] pypy fix-result-types: Let W_Ufunc2 handle scalars in the same way as W_Ufunc1
rlamy
- [pypy-commit] pypy fix-result-types: Create W_Ufunc2.find_specialization()
rlamy
- [pypy-commit] pypy numpy-fixes: test, fix for calling __array_finalize__ on set_shape (needed for subtypes)
mattip
- [pypy-commit] pypy pythonoptimize-env: close branch to be merged
mattip
- [pypy-commit] pypy default: merge pythonoptimize-env fix issue #2004 by relating to PYTHONOPTIMIZE
mattip
- [pypy-commit] cffi cffi-1.0: Add RTLD_xxx flags to the CompiledFFI type
arigo
- [pypy-commit] cffi cffi-1.0: writing docs, in-progress
arigo
- [pypy-commit] pypy numpy-fixes: close merged branch
mattip
- [pypy-commit] pypy default: merge newest changes to default
mattip
- [pypy-commit] cffi cffi-1.0: Best-effort attempt at supporting C++. There is still one issue shown
arigo
- [pypy-commit] cffi release-0.9: hg merge default
arigo
- [pypy-commit] cffi cffi-1.0: Finish the docs
arigo
- [pypy-commit] cffi cffi-1.0: point inside the doc
arigo
- [pypy-commit] cffi cffi-1.0: Use execfile() to load the build script from setuptools_ext, instead
arigo
- [pypy-commit] cffi cffi-1.0: Last TODO item: setuptools on out-of-line ABI-mode modules
arigo
- [pypy-commit] cffi cffi-1.0: small doc fixes
arigo
- [pypy-commit] cffi cffi-1.0: Describe how to write in a way that loads both in cffi-0.9 and cffi-1.0
arigo
- [pypy-commit] cffi cffi-1.0: Naive. This has more chances to work...
arigo
- [pypy-commit] cffi cffi-1.0: or like that
arigo
- [pypy-commit] cffi cffi-1.0: (kostia) clarify
arigo
- [pypy-commit] cffi cffi-1.0: Expand a bit the setup.py compatibility section
arigo
- [pypy-commit] cffi cffi-1.0: typo
arigo
- [pypy-commit] cffi cffi-1.0: Mention this
arigo
- [pypy-commit] cffi cffi-1.0: Last fix, now test the C++ mode
arigo
- [pypy-commit] cffi cffi-1.0: Explicit
arigo
- [pypy-commit] pypy default: fix this test (fails in x86/test/test_fficall.py)
arigo
- [pypy-commit] pypy cffi-1.0: First test passes
arigo
- [pypy-commit] pypy fix-result-types: push more logic inside allowed_types()
rlamy
- [pypy-commit] pypy cffi-1.0: fixes (test_recompiler.py)
arigo
- [pypy-commit] pypy cffi-1.0: test_global_var
arigo
- [pypy-commit] pypy cffi-1.0: test_function
arigo
- [pypy-commit] pypy cffi-1.0: test_dlclose
arigo
- [pypy-commit] pypy cffi-1.0: Structs and unions
arigo
- [pypy-commit] pypy cffi-1.0: test_enum
arigo
- [pypy-commit] pypy cffi-1.0: test_*_struct
arigo
- [pypy-commit] pypy cffi-1.0: test_include_1
arigo
- [pypy-commit] pypy cffi-1.0: bug fixes
arigo
- [pypy-commit] pypy cffi-1.0: move the hack
arigo
- [pypy-commit] cffi cffi-1.0: Generalize: in pypy it turns out we can naturally give a more precise message
arigo
- [pypy-commit] cffi cffi-1.0: Update version number here
arigo
- [pypy-commit] pypy cffi-1.0: Add a test and fix
arigo
- [pypy-commit] cffi cffi-1.0: Initialize the __name__ and __file__ arguments when we're about to
arigo
- [pypy-commit] cffi cffi-1.0: One more pass through this doc
arigo
- [pypy-commit] cffi cffi-1.0: rewrites
arigo
- [pypy-commit] pypy vecopt: improved the scheduling (missed to emit pack/unpack ops), work in progress
plan_rich
- [pypy-commit] cffi cffi-1.0: Forgot about the version checking and specifying in the ABI-mode modules
arigo
- [pypy-commit] pypy cffi-1.0: Update to cffi/2d4469c2fec5
arigo
- [pypy-commit] pypy cffi-1.0: Update the importing tool
arigo
- [pypy-commit] pypy cffi-1.0: Import cffi-1.0/2d4469c2fec5
arigo
- [pypy-commit] cffi cffi-1.0: Close branch, ready for merge
arigo
- [pypy-commit] cffi default: hg merge cffi-1.0
arigo
- [pypy-commit] cffi default: Add a mostly empty "whatsnew.rst"
arigo
- [pypy-commit] cffi default: Nicer to have these two paragraphs listed here too, even though they are from the same page
arigo
- [pypy-commit] pypy numpy-flags: add attribute flags to BaseConcreteArray
mattip
- [pypy-commit] pypy numpy-flags: pass tests in test_flagobj, add test for str(flags)
mattip
- [pypy-commit] pypy numpy-flags: add more properties
mattip
- [pypy-commit] pypy numpy-flags: cleanup, override __repr__
mattip
- [pypy-commit] pypy more-rposix: Restore dont_look_inside annotation
amauryfa
- [pypy-commit] pypy numpy-flags: document branch
mattip
- [pypy-commit] pypy numpy-flags: close branch to be merged
mattip
- [pypy-commit] pypy default: merge numpy-flags which completes the ndarray.flags property
mattip
- [pypy-commit] pypy default: rework failing tests
mattip
- [pypy-commit] cffi default: Mention two MSVC compilers
arigo
- [pypy-commit] cffi default: Skip this test on pypy (it's done already pre-translated)
arigo
- [pypy-commit] pypy cffi-1.0: Make the cffi tests pass on a translated pypy
arigo
- [pypy-commit] pypy fix-result-types: don't use find_binop_result_dtype() in W_Ufunc2.call()
rlamy
- [pypy-commit] pypy fix-result-types: fix for comparison ufuncs
rlamy
- [pypy-commit] stmgc default: re-add some missing comments in stmgc.h
Raemi
- [pypy-commit] pypy default: for consistency, the in-place operators need the same bug compatibility hack as the regular operators
arigo
- [pypy-commit] pypy vecopt: removed a bug where packtype was modified but not copied before that
plan_rich
- [pypy-commit] pypy vecopt: cvtpd2dq packs ints to the lower quadword. that is why it did not work, starting to rethink the conversion function
plan_rich
- [pypy-commit] pypy vecopt: finished the unpacking for float32/64 and int32/64. added x86 packed mul operations (don't know if we ever can use them for int64)
plan_rich
- [pypy-commit] pypy vecopt: removed annotation errors that occur when building the jit
plan_rich
- [pypy-commit] cffi release-1.0: Make the release branch
arigo
- [pypy-commit] cffi release-1.0: add the md5/sha1
arigo
- [pypy-commit] pypy fix-result-types: fix
rlamy
- [pypy-commit] pypy fix-result-types: Don't use find_unaryop_result_dtype() in W_Ufunc.reduce()
rlamy
- [pypy-commit] cffi default: Add some more "codeblock:: python"
arigo
- [pypy-commit] pypy cffi-1.0: Removing "build" dirs from the tests
arigo
- [pypy-commit] pypy cffi-1.0: close branch, ready to merge
arigo
- [pypy-commit] pypy default: hg merge cffi-1.0
arigo
- [pypy-commit] pypy fix-result-types: Create np.promote_types(), with failing tests
rlamy
- [pypy-commit] cffi default: Add a small warning
arigo
- [pypy-commit] cffi default: Ah, we don't have a _static dir
arigo
- [pypy-commit] pypy fix-result-types: kill unused flags in find_binop_result_dtype()
rlamy
- [pypy-commit] pypy fix-result-types: kill unused find_unaryop_result_dtype()
rlamy
- [pypy-commit] pypy default: Adapt package.py for the out-of-line modules. Currently there is only one, for gdbm.
arigo
- [pypy-commit] pypy default: These functions can be static too. It's certainly better than exporting (and potentially conflicting)
arigo
- [pypy-commit] pypy default: audioop: use an out-of-line module
arigo
- [pypy-commit] pypy default: syslog
arigo
- [pypy-commit] pypy default: _curses_cffi
arigo
- [pypy-commit] pypy default: _tkinter as an out-of-line module
arigo
- [pypy-commit] pypy fix-result-types: Create static promotion_table and use it in np.promote_types()
rlamy
- [pypy-commit] cffi default: Mention that lib.__dict__ doesn't exist any more, too
arigo
- [pypy-commit] pypy default: _sqlite3_cffi
arigo
- [pypy-commit] pypy default: Skip some of these tests in case of runappdirect (they are also in ../test_pypy_c/cffi_tests)
arigo
- [pypy-commit] pypy default: Write a skipped test as documentation
arigo
- [pypy-commit] pypy default: _pwd_cffi, to make it a bit more portable than the ctypes version
arigo
- [pypy-commit] pypy default: move grp.py's logic inside _pwdgrp_build.py too
arigo
- [pypy-commit] pypy default: test, fix for str(box('abc')) => 'abc' rather than "'abc'"
mattip
- [pypy-commit] pypy fix-result-types: precompute string length for integer itemtypes
rlamy
- [pypy-commit] pypy fix-result-types: fix promote_types(<numeric>, <string>)
rlamy
- [pypy-commit] pypy fix-result-types: update test for correct numpy 1.9 behaviour
rlamy
- [pypy-commit] cffi default: move up this warning
arigo
- [pypy-commit] cffi default: Test and fix (reported by sarvi)
arigo
- [pypy-commit] pypy default: Fix whatsnew
arigo
- [pypy-commit] pypy default: Fix for cpyext's test_ztranslation: don't import _cffi_backend from the
arigo
- [pypy-commit] pypy default: Remove this file: it could be done by running a generic test on top
arigo
- [pypy-commit] pypy default: The test can't write to the 'lib' object (it is now read-only). Tweak monkey-patching.
arigo
- [pypy-commit] cffi default: Issue #193: if we use a struct between the first cdef() where it is
arigo
- [pypy-commit] cffi default: Write this bug fix into whatsnew.rst.
arigo
- [pypy-commit] cffi default: Mention a0f4efcdb20b
arigo
- [pypy-commit] pypy vecopt: helping the rtyper
plan_rich
- [pypy-commit] pypy vecopt: excluded getfield gc/raw from the load tracking. there is no index on this operation. index is in the descriptor
plan_rich
- [pypy-commit] pypy vecopt: turned off vectorize opt for all jit drivers and enabled vectorize opt in micronumpy loop jit drivers
plan_rich
- [pypy-commit] pypy vecopt: added debug output in the vecopt method and log it in the logfile
plan_rich
- [pypy-commit] pypy.org extradoc: Update the "packaging" section with separate pre-cffi-1.0 and
arigo
- [pypy-commit] pypy py3.3: Replace tabs by spaces in lib_pypy/_lzma.py.
mjacob
- [pypy-commit] cffi default: Windows: pass along the result (success/error) of FreeLibrary
arigo
- [pypy-commit] cffi default: clarify
arigo
- [pypy-commit] cffi default: tweak tweak the docs
arigo
- [pypy-commit] pypy default: BogusPureField -> BogusImmutableField
cfbolz
- [pypy-commit] pypy pypyw: test, add *w.exe target when shared build for win32. Test is still failing.
mattip
- [pypy-commit] pypy fix-result-types: prepare find_result_type() for implementing correct handling of scalars
rlamy
- [pypy-commit] pypy win32-optionals: throw away branch
mattip
- [pypy-commit] pypy closed-branches: merge dead branch
mattip
- [pypy-commit] cffi fix-multiple-sources-arguments: Fix multiple sources arguments passed to setuptools.
chrippa
- [pypy-commit] cffi default: Merged in chrippa/cffi/fix-multiple-sources-arguments (pull request #60)
arigo
- [pypy-commit] cffi default: Mention 33a80cca1c69.
arigo
- [pypy-commit] cffi default: hg merge release-1.0
arigo
- [pypy-commit] pypy fold-arith-ops: add test for removing multiple adds
Berkin Ilbeyi
- [pypy-commit] pypy fold-arith-ops: more/better testing for removing multiple int_add's
Berkin Ilbeyi
- [pypy-commit] pypy fold-arith-ops: remove multiple adds on add chains ("1 + 1 + 1 + ...")
Berkin Ilbeyi
- [pypy-commit] pypy fold-arith-ops: minor change in test for removing multiple adds
Berkin Ilbeyi
- [pypy-commit] pypy fold-arith-ops: check if the addition is a valid int before folding int_add's
Berkin Ilbeyi
- [pypy-commit] pypy py3k: hg merge da90c30dc0dd
mjacob
- [pypy-commit] pypy py3k: Remove test which doesn't make sense on py3k.
mjacob
- [pypy-commit] pypy py3k: hg merge 8cb5c941efb4
mjacob
- [pypy-commit] pypy py3k: hg merge
mjacob
- [pypy-commit] cffi default: Bump the version to 1.0.1
arigo
- [pypy-commit] cffi default: Add an example purely for performance
arigo
- [pypy-commit] pypy default: Bump the version to 1.0.1
arigo
- [pypy-commit] pypy fold-arith-ops: reduce code duplication by folding the two cases into one
cfbolz
- [pypy-commit] pypy fold-arith-ops: run less code in a try...except block
cfbolz
- [pypy-commit] cffi release-1.0: hg merge default
arigo
- [pypy-commit] cffi release-1.0: MD5/SHA1 for this release, which is cffi 1.0.1
arigo
- [pypy-commit] pypy default: update this title
arigo
- [pypy-commit] pypy default: Update the cffi pure Python package to 1.0.1
arigo
- [pypy-commit] pypy fold-arith-ops: remove more code duplication, add comments
Berkin Ilbeyi
- [pypy-commit] pypy default: fix
arigo
- [pypy-commit] pypy vecopt: refactored the dependency construction for guards
plan_rich
- [pypy-commit] pypy vecopt: removed attribute error when printing unrolled/vectorized loop
plan_rich
- [pypy-commit] pypy vecopt: rewritten the guard strengthening. it is now independent from vecopt (still contained in the same file though).
plan_rich
- [pypy-commit] pypy fold-arith-ops: when folding int_add's, handle overflow correctly
Berkin Ilbeyi
- [pypy-commit] pypy fold-arith-ops: change add/sub elim test to reflect int_add folding
Berkin Ilbeyi
- [pypy-commit] pypy fix-result-types: Add num2dtype() helper
rlamy
- [pypy-commit] pypy fix-result-types: Implement numpy's complicated scalar handling rules in result_type()
rlamy
- [pypy-commit] pypy fix-result-types: test np.result_type(); fix
rlamy
- [pypy-commit] pypy fix-result-types: Clean up code in casting.py
rlamy
- [pypy-commit] pypy pypyw: fix standalone test, add pypyw.exe to driver and test
mattip
- [pypy-commit] pypy pypyw: merge default into branch
mattip
- [pypy-commit] pypy pypyw: package pypyw, fix cffi module build for obscure Python27.lib location
mattip
- [pypy-commit] pypy pypyw: document branch
mattip
- [pypy-commit] cffi default: Test and fix: ABI out-of-line didn't support C functions with
arigo
- [pypy-commit] cffi default: Bump the version number to 1.0.2
arigo
- [pypy-commit] pypy default: Update to cffi 1.0.2
arigo
- [pypy-commit] cffi default: Add doc
arigo
- [pypy-commit] cffi default: Defines the macro _CFFI_ before including Python.h.
arigo
- [pypy-commit] pypy default: Windows: when _CFFI_ is defined, don't link with python27.lib at all.
arigo
- [pypy-commit] pypy default: import cffi 1.0.2 again
arigo
- [pypy-commit] cffi default: Fix comments
arigo
- [pypy-commit] cffi default: Expand the problem
arigo
- [pypy-commit] pypy fix-result-types: Handle record dtypes in can_cast()
rlamy
- [pypy-commit] pypy fix-result-types: Use find_result_type in concatenate()
rlamy
- [pypy-commit] pypy fix-result-types: Create scalar2dtype() to simplify find_dtype_for_scalar()
rlamy
- [pypy-commit] pypy fix-result-types: kill find_dtype_for_scalar()
rlamy
- [pypy-commit] pypy fix-result-types: improve compatibility of error messages
rlamy
- [pypy-commit] pypy pypyw: remove hack made obsolete by 2bcf56036762
mattip
- [pypy-commit] pypy pypyw: close branch to be merged
mattip
- [pypy-commit] pypy default: merge pypyw which provides pypyw.exe as well as pypy.exe on win32
mattip
- [pypy-commit] pypy default: Add all the "--without-NAME" options for all cffi build scripts
arigo
- [pypy-commit] cffi default: Pull request #61 fixing issue #182 (slightly edited):
arigo
- [pypy-commit] cffi default: On PyPy, cffi is preinstalled and it is not possible, at least for now,
arigo
- [pypy-commit] pypy object-dtype:
mattip
- [pypy-commit] pypy quieter-translation:
mattip
- [pypy-commit] pypy closed-branches: merge object-dtype into closed-branches
mattip
- [pypy-commit] pypy closed-branches: merge quieter-translation into closed-branches
mattip
- [pypy-commit] pypy release-2.6.x: start release cycle
mattip
- [pypy-commit] pypy default: bump version number, unsure whether pypy-2.7.0 is such a good idea
mattip
- [pypy-commit] pypy default: move whatsnew-head.rst
mattip
- [pypy-commit] pypy default: re-add whatsnew-head.rst
mattip
- [pypy-commit] pypy release-2.6.x: merge default into branch
mattip
- [pypy-commit] pypy release-2.6.x: work on release announcement
mattip
- [pypy-commit] pypy default: update to exactly 1.0.2
arigo
- [pypy-commit] pypy fold-arith-ops: document branch
mattip
- [pypy-commit] pypy fold-arith-ops: close branch to be merged
mattip
- [pypy-commit] pypy default: merge fold-arith-ops which removes multiple adds on add chains ("1 + 1 + 1 + ...")
mattip
- [pypy-commit] pypy default: dummy commit for whatsnew-head
mattip
- [pypy-commit] pypy default: update whatsnew-head
mattip
- [pypy-commit] pypy release-2.6.x: merge default into release
mattip
- [pypy-commit] pypy release-2.6.x: tweaks
mattip
- [pypy-commit] pypy release-2.6.x: more links
mattip
- [pypy-commit] pypy vecopt: resolved problem with guard strengthening (boolinverse needed if guard_false)
plan_rich
- [pypy-commit] pypy release-2.6.x: fix links, add freebsd (ebarret)
mattip
- [pypy-commit] cffi default: typo
arigo
- [pypy-commit] cffi default: (kostialopuhin, pull request)
arigo
- [pypy-commit] cffi osx-tests: add missing import for OS X
kostialopuhin
- [pypy-commit] cffi default: hg merge osx-tests
arigo
- [pypy-commit] pypy default: More tweaks (thanks Yichao Yu commenting on 98e7823b191a)
arigo
- [pypy-commit] pypy default: Issue #2035
arigo
- [pypy-commit] pypy exp-with-blackhole: try to dump info
fijal
- [pypy-commit] pypy exp-with-blackhole: fix rpython
fijal
- [pypy-commit] pypy exp-with-blackhole: oops
fijal
- [pypy-commit] pypy exp-with-blackhole: more rpython fixes, print pendingfields
fijal
- [pypy-commit] pypy exp-with-blackhole: close experiment branch
fijal
- [pypy-commit] pypy release-2.6.x: rework the release announcement, I will make sure that vmprof works on 2.6
fijal
- [pypy-commit] cffi default: Update to current
arigo
- [pypy-commit] pypy default: try a bit harder to load vmprof at run time
fijal
- [pypy-commit] pypy fix-result-types: return the right int32 dtype instance on 32-bit
rlamy
- [pypy-commit] pypy fix-result-types: Use promote_types() for binary ufunc resolution in some cases
rlamy
- [pypy-commit] stmgc use-gcc: A branch in which to use a patched gcc (for the extra feature)
arigo
- [pypy-commit] stmgc use-gcc: in-progress
arigo
- [pypy-commit] stmgc use-gcc: Add the gcc patch
arigo
- [pypy-commit] cffi release-1.0: fix test failure on Win32 - a function need to be exported to be callable.
grizzlynyo
- [pypy-commit] cffi default: Update
arigo
- [pypy-commit] stmgc use-gcc: Avoid 'uint8_t' as fields because gcc considers that read/writes to it
arigo
- [pypy-commit] stmgc use-gcc: meh
arigo
- [pypy-commit] pypy default: hard to test case of error
fijal
- [pypy-commit] stmgc use-gcc: updates
arigo
- [pypy-commit] stmgc use-gcc: improve error message on a non-patched gcc
arigo
- [pypy-commit] stmgc use-gcc: Expect the patched version of gcc to be available under the name 'gcc-seg-gs'.
arigo
- [pypy-commit] stmgc use-gcc: merge heads
arigo
- [pypy-commit] stmgc use-gcc: Move and update the patch
arigo
- [pypy-commit] pypy stmgc-c7: import stmgc/c7 use-gcc, and use 'gcc-seg-gs'
arigo
- [pypy-commit] stmgc use-gcc: Add a README
arigo
- [pypy-commit] stmgc use-gcc: Expand the instructions as I found it necessary on some other system.
arigo
- [pypy-commit] stmgc use-gcc: Add the "__seg_fs" address space too, for completeness.
arigo
- [pypy-commit] stmgc use-gcc: tweaks
arigo
- [pypy-commit] stmgc use-gcc: more
arigo
- [pypy-commit] stmgc use-gcc: fix test
arigo
- [pypy-commit] stmgc use-gcc: Give up and document the gcc-only restriction
arigo
- [pypy-commit] pypy fix-result-types: In ufunc.accumulate(), the output dtype isn't necessarily the same as calc_dtype
rlamy
- [pypy-commit] buildbot default: add dw's slave
mattip
- [pypy-commit] buildbot default: hopefully clarify
mattip
- [pypy-commit] pypy fix-result-types: Add a test for the complicated relations between scalar integer types (verified on upstream numpy 1.9.2)
rlamy
- [pypy-commit] pypy fix-result-types: Set np.intp/np.uintp correctly on 32-bit platforms
rlamy
- [pypy-commit] pypy fix-result-types: oops
rlamy
- [pypy-commit] pypy closed-branches: Merge closed head 4835eb5bf06a on branch pytest-25
mattip
- [pypy-commit] pypy closed-branches: Merge closed head 3d26064d218e on branch no-write-barrier-in-const-ptrs
mattip
- [pypy-commit] pypy closed-branches: Merge closed head 6c50bbbada5b on branch shadowstack-again
mattip
- [pypy-commit] pypy closed-branches: Merge closed head 6d312a8d75bf on branch rtyper-stuff
mattip
- [pypy-commit] pypy closed-branches: Merge closed head c6314683ba8d on branch var-in-Some
mattip
- [pypy-commit] pypy closed-branches: Merge closed head 0a65021c50aa on branch ClassRepr
mattip
- [pypy-commit] pypy closed-branches: Merge closed head c3e9466adc06 on branch expressions
mattip
- [pypy-commit] pypy closed-branches: Merge closed head 76c31595eda4 on branch exp-with-blackhole
mattip
- [pypy-commit] pypy closed-branches: re-close this branch
mattip
- [pypy-commit] pypy default: update contributors with 9 new names
mattip
- [pypy-commit] pypy default: of course
fijal
- [pypy-commit] pypy default: don't close the handle, makes symbols invalid
fijal
- [pypy-commit] pypy default: we fixed the obscure .so mess
fijal
- [pypy-commit] pypy optresult: fix a small irrelevant test to get started
fijal
- [pypy-commit] pypy optresult: fix bound issue and fix pure issue with input args or consts
fijal
- [pypy-commit] pypy optresult: improve the situation for some arithmetic operations
fijal
- [pypy-commit] pypy optresult: hooray passed more of the tests than failed
fijal
- [pypy-commit] pypy optresult: one more test
fijal
- [pypy-commit] pypy optresult: whack a bit at those tests
fijal
- [pypy-commit] pypy optresult: register dirty fields
fijal
- [pypy-commit] pypy optresult: record the dirty field here
fijal
- [pypy-commit] pypy optresult: use an ordered dictionary here (but we might as well use a value on the descr at some point)
fijal
- [pypy-commit] pypy optresult: fix forcing before guards
fijal
- [pypy-commit] pypy optresult: disable handling pendingfields in resume, there is zero chance we can fix it
fijal
- [pypy-commit] pypy vecopt: removed the const arg for each vecop (but not load)
plan_rich
- [pypy-commit] pypy release-2.6.x: polish the release announcement
mattip
- [pypy-commit] pypy vecopt: removed rpython problems
plan_rich
- [pypy-commit] pypy default: fix win32 errors
mattip
- [pypy-commit] cffi default: Bump the version number to 1.0.3
arigo
- [pypy-commit] cffi release-1.0: What's New: nothing, basically
arigo
- [pypy-commit] cffi release-1.0: md5/sha1
arigo
- [pypy-commit] pypy default: Update to cffi 1.0.3.
arigo
- [pypy-commit] pypy default: skip these on win32
mattip
- [pypy-commit] pypy default: fix skip message
arigo
- [pypy-commit] pypy default: update to cffi 1.0.3
arigo
- [pypy-commit] pypy fix-result-types: fix 32-bit dtype tests and remove duplication with test_scalar.py
rlamy
- [pypy-commit] pypy default: Fix the test by adding a _fake option that doesn't rely on a pypy-c and
arigo
- [pypy-commit] pypy release-2.6.x: formatting, mention new contributors
mattip
- [pypy-commit] pypy default: simplify
arigo
- [pypy-commit] cffi default: Silence some warnings with gcc 5.1: if T is a type smaller than 'int',
arigo
- [pypy-commit] pypy optresult: start fighting with heapcache of arrays
fijal
- [pypy-commit] pypy optresult: fixes
fijal
- [pypy-commit] pypy optresult: oops forgot about this one
fijal
- [pypy-commit] pypy vecopt: added test case for float_truediv (was not included in transformation), float_neg, float_abs
plan_rich
- [pypy-commit] pypy vecopt: added float_neg and float_abs implementations to x86
plan_rich
- [pypy-commit] pypy vecopt: removed count from float_expand asm args, added float expand for single float (parts of it in previous commit)
plan_rich
- [pypy-commit] pypy vecopt: constant expansion in trace now adds the constants to the heap memory (instead of crashing)
plan_rich
- [pypy-commit] pypy vecopt: rpy test passes for recent changes
plan_rich
- [pypy-commit] pypy optresult: minor fixes, pass more tests about arrayitems
fijal
- [pypy-commit] pypy optresult: start passing the merge-guards kind of tests
fijal
- [pypy-commit] pypy optresult: too eager assert
fijal
- [pypy-commit] pypy optresult: finish the deal with merging guards
fijal
- [pypy-commit] pypy optresult: seems we're smarter these days
fijal
- [pypy-commit] cffi default: theoretical fix for pypy3
arigo
- [pypy-commit] pypy optresult: unify some more
fijal
- [pypy-commit] pypy optresult: whack whack whack
fijal
- [pypy-commit] cffi default: Performance: no real need to call PyArg_ParseTuple() here
arigo
- [pypy-commit] pypy default: Bumped greenlet version to match latest on pypi (no new features)
alex_gaynor
- [pypy-commit] pypy default: Tweak: change 'structwrapper' to be used as a wrapper for all built-in
arigo
- [pypy-commit] pypy optresult: hack enough to start passing resume tests
fijal
- [pypy-commit] cffi default: Comment
arigo
- [pypy-commit] pypy optresult: fix constant aliasing
fijal
- [pypy-commit] pypy optresult: fix forcing, but the complexity is a bit wrong
fijal
- [pypy-commit] pypy optresult: whack at vrefs
fijal
- [pypy-commit] pypy default: Add this to the list of symbols to rename from PyXxx to PyPyXxx.
arigo
- [pypy-commit] pypy optresult: finish with vref I hope
fijal
- [pypy-commit] pypy optresult: first go at arraycopy
fijal
- [pypy-commit] pypy optresult: a few cheap wins
fijal
- [pypy-commit] cffi default: ffi.addressof(lib, "function") now returns a regular cdata function pointer
arigo
- [pypy-commit] pypy vecopt: enhanced test env to check schedule operation
plan_rich
- [pypy-commit] pypy vecopt: added missing result for pack instruction (expand box to vector)
plan_rich
- [pypy-commit] pypy default: Test and fix: the test (s == "") was optimized by incorrectly assuming
arigo
- [pypy-commit] pypy optresult: basic strlen opts
fijal
- [pypy-commit] pypy default: A tentative fix for an issue we're unlikely to be able to reproduce,
arigo
- [pypy-commit] pypy default: Fix on OS/X: match the behavior of CPython.
arigo
- [pypy-commit] cffi default: Bump version number
arigo
- [pypy-commit] cffi default: Document
arigo
- [pypy-commit] pypy default: Update: ffi.addressof(lib, "funcname")
arigo
- [pypy-commit] pypy optresult: start whacking at vstring
fijal
- [pypy-commit] pypy optresult: copy-paste some more methods into VStringPlainInfo
fijal
- [pypy-commit] pypy optresult: hack on a const a bit
fijal
- [pypy-commit] pypy default: Fixed some string formatting in a sqlite3 exception
alex_gaynor
- [pypy-commit] pypy optresult: more hacking on vstring
fijal
- [pypy-commit] pypy optresult: enough to pass the first string equality test
fijal
- [pypy-commit] pypy default: Added tag release-2.6.0 for changeset fcdb94156515
mattip
- [pypy-commit] benchmarks default: Under some unknown circumstance, it is possible to hit a path where a Branch instance
arigo
- [pypy-commit] pypy fix-result-types: precompute W_Ufunc1.allowed_types()
rlamy
- [pypy-commit] pypy fix-result-types: precompute W_Ufunc2.allowed_types()
rlamy
- [pypy-commit] pypy fix-result-types: Fix dtype resolution in ufunc.reduce()
rlamy
- [pypy-commit] pypy optresult: small fix to the level1 equality
fijal
- [pypy-commit] pypy optresult: two more tests
fijal
- [pypy-commit] pypy default: release management tweaks
mattip
- [pypy-commit] pypy optresult: pass the first test of slices
fijal
- [pypy-commit] pypy optresult: blind fixes on str equality
fijal
- [pypy-commit] pypy optresult: str equality does not mean args are non-null
fijal
- [pypy-commit] pypy optresult: handle null in a few more places
fijal
- [pypy-commit] pypy optresult: few small fixes
fijal
- [pypy-commit] pypy optresult: kill some dead code
fijal
- [pypy-commit] pypy optresult: more unused code
fijal
- [pypy-commit] pypy optresult: fix failures on pypy and also skip the test that's a bit hopeless right now
fijal
- [pypy-commit] pypy optresult: fix the last problem
fijal
- [pypy-commit] pypy optresult: comment out pieces of unrolling, we have to think what to do next
fijal
- [pypy-commit] pypy optresult: enough to pass test_loop.py
fijal
- [pypy-commit] pypy optresult: TODO that's not tested by test_optimizebasic
fijal
- [pypy-commit] pypy optresult: fix test_runner for changes in new_with_vtable
fijal
- [pypy-commit] pypy vecopt: added a vectorize_user parameter to try the vectorization on user code (array[i], numpy.array[i]), removed some old checks for test_axissum
plan_rich
- [pypy-commit] pypy optresult: simple whacking at the backend
fijal
- [pypy-commit] pypy optresult: start passing backend tests
fijal
- [pypy-commit] pypy optresult: pass some of the backend tests
fijal
- [pypy-commit] pypy optresult: start fighting with rewriting
fijal
- [pypy-commit] cffi default: Test and various fixes for the location of the produced files, notably
arigo
- [pypy-commit] cffi default: Fix the test on Windows
arigo
- [pypy-commit] cffi default: Issue #197: this import fails for some people on CPython>=3.3 without
arigo
- [pypy-commit] cffi default: Mention this too
arigo
- [pypy-commit] pypy default: import cffi/1f18df199d5b
arigo
- [pypy-commit] pypy default: Add missing file
arigo
- [pypy-commit] pypy optresult: fight with rewriting quite a bit more
fijal
- [pypy-commit] pypy default: import cffi/c0ca172d6ce6
arigo
- [pypy-commit] cffi default: Issue 198: corrupted constant of type 'struct'
arigo
- [pypy-commit] pypy default: update to cffi/1283fea71028 for issue 198.
arigo
- [pypy-commit] cffi default: Mention issue #198
arigo
- [pypy-commit] pypy vecopt: factored out a rename object for trace operations
plan_rich
- [pypy-commit] pypy vecopt: retinkering the dependency construction, statements with sideeffects need stronger dependencies
plan_rich
- [pypy-commit] cffi default: Add a passing test
arigo
- [pypy-commit] cffi default: Issue #198 bis: fix for constants of unknown size
arigo
- [pypy-commit] pypy default: hack around a stupid limitation, now _vmprof is not enabled when JIT is not on
fijal
- [pypy-commit] cffi default: more about issue 198: Test and fix
arigo
- [pypy-commit] pypy default: do it yet differently
fijal
- [pypy-commit] pypy default: Add passing tests
arigo
- [pypy-commit] pypy default: we no longer need it
fijal
- [pypy-commit] pypy optresult: finish fixing rewrite
fijal
- [pypy-commit] pypy optresult: minor hacks until we encounter an actual obstacle
fijal
- [pypy-commit] pypy optresult: whack whack whack until we get somewhere with the backend tests
fijal
- [pypy-commit] pypy optresult: fix those tests
fijal
- [pypy-commit] pypy default: Moved tag release-2.6.0 to changeset e03971291f3a (from changeset fcdb94156515)
mattip
- [pypy-commit] pypy vecopt: correctly emitting normal guards not subject of the strengh opt
plan_rich
- [pypy-commit] pypy optresult: fix some tests
fijal
- [pypy-commit] pypy fix-result-types: Ensure that casting_table and promotion_table are built deterministically
rlamy
- [pypy-commit] pypy vecopt: uncommented more tests to see how well current version works
plan_rich
- [pypy-commit] cffi default: ABI mode: allow constants of any type, which are looked up from the lib
arigo
- [pypy-commit] pypy default: update to cffi/4f05d11fa6ab
arigo
- [pypy-commit] pypy default: Update to cffi/22eff74bf6e8
arigo
- [pypy-commit] pypy default: add a passing test (cffi/020ef1915196)
arigo
- [pypy-commit] pypy default: pypy/tool/import_cffi: import revision 020ef1915196
arigo
- [pypy-commit] pypy py3k: Fix stack depth computation.
mjacob
- [pypy-commit] pypy fix-result-types: Rename promote_types() to w_promote_types() and _promote_types() to promote_types()
rlamy
- [pypy-commit] pypy fix-result-types: Test promote_types() only at app-level, for easier verification of test correctness
rlamy
- [pypy-commit] cffi default: py3 fix
arigo
- [pypy-commit] cffi default: Better compatibility: detect if we have an old cffi running a new ABI
arigo
- [pypy-commit] pypy default: update to cffi/4d3306c3afcc
arigo
- [pypy-commit] pypy py3k: hg merge 5acade5a80c5
mjacob
- [pypy-commit] pypy py3k: Remove unnecessary import (which also breaks translation on py3k).
mjacob
- [pypy-commit] pypy py3k: 2to3: There is no space.w_long / space.w_memoryview in Py3k.
mjacob
- [pypy-commit] pypy fix-result-types: Move casting_table and promotion_table to casting.py
rlamy
- [pypy-commit] pypy fix-result-types: Move can_cast_to() and can_cast_itemtype() to casting.py
rlamy
- [pypy-commit] pypy fix-result-types: Move find_dtype_for_seq() to ctors.py
rlamy
- [pypy-commit] pypy fix-result-types: fix test
rlamy
- [pypy-commit] pypy fix-result-types: fix for -A testing
mattip
- [pypy-commit] pypy fix-result-types: test, fix scalar indexing for -A compatability
mattip
- [pypy-commit] pypy fix-result-types: fix tests, error msg format for -A compatability
mattip
- [pypy-commit] pypy default: Add missing operation
arigo
- [pypy-commit] stmgc use-gcc: Add missing docs
arigo
- [pypy-commit] pypy vecopt: bug fix. did not copy descriptors while unrolling instructions, simplifcations and use copy_all_attr of the descr instead of doing it manually
plan_rich
- [pypy-commit] pypy optresult: fix last test
fijal
- [pypy-commit] pypy optresult: kill infos on operations. fix failargs while in rewrite
fijal
- [pypy-commit] pypy optresult: sort of try to disable unrolling
fijal
- [pypy-commit] pypy optresult: fix fix fix
fijal
- [pypy-commit] pypy default: Fix the keyword argument names that differ from CPython
arigo
- [pypy-commit] pypy optresult: small fixes
fijal
- [pypy-commit] pypy vecopt: generalized extend
plan_rich
- [pypy-commit] pypy vecopt: first try to merge iterators in call2
plan_rich
- [pypy-commit] pypy default: Document this as a (now-)known difference with CPython. Thanks
arigo
- [pypy-commit] pypy optresult: fix copying value in one more place
fijal
- [pypy-commit] pypy optresult: skip some more tests
fijal
- [pypy-commit] pypy optresult: fix pending_setfields for arrays
fijal
- [pypy-commit] stmgc use-gcc: Close branch ready to merge
arigo
- [pypy-commit] stmgc default: hg merge use-gcc
arigo
- [pypy-commit] pypy fix-result-types: Fix dtype promotion to unicode
rlamy
- [pypy-commit] pypy fix-result-types: Fix inplace operations on arrays in cases where an unsafe cast is needed
rlamy
- [pypy-commit] pypy fix-result-types: Close branch before merging
rlamy
- [pypy-commit] pypy default: Merge branch 'fix-result-types'
rlamy
- [pypy-commit] pypy default: update whatsnew
rlamy
- [pypy-commit] stmgc default: Backing out fba24ba1a75f: gcc does some optimizations before entering
arigo
- [pypy-commit] pypy stmgc-c7: Backed out changeset 4d93e7a10f91: gcc does some optimizations before
arigo
- [pypy-commit] pypy.org extradoc: Put the "portable binaries" link more to the front
arigo
- [pypy-commit] pypy.org extradoc: Give a more precise link, as it seems to confuse people
arigo
- [pypy-commit] cffi default: Issue 75: implement multidimensional use of '[...]'.
arigo
- [pypy-commit] cffi default: Issues 69, 73: add the syntax "typedef int... foo_t; ".
arigo
- [pypy-commit] cffi default: Issue 200: bad interaction between "ffi.typeof(function_t)" and
arigo
- [pypy-commit] pypy default: import cffi/83a960dee7ad
arigo
- [pypy-commit] cffi default: Tweaks and minor fixes
arigo
- [pypy-commit] pypy default: update to cffi/3c02643ecb69
arigo
- [pypy-commit] cffi default: pypy compat
arigo
- [pypy-commit] pypy default: Fix test_lloperation
arigo
- [pypy-commit] pypy default: import cffi/a0a5812b5d41
arigo
- [pypy-commit] cffi default: Check that the types declared with "typedef int... foo_t; " are really
arigo
- [pypy-commit] pypy default: import cffi/1e7ca6f81db8
arigo
- [pypy-commit] cffi default: Use '...' here
arigo
- [pypy-commit] pypy default: Use the new syntax "typedef int... foo_t" where appropriate
arigo
- [pypy-commit] cffi default: Precision
arigo
- [pypy-commit] pypy default: Issue 2047: add a note
arigo
- [pypy-commit] cffi default: The next release will be 1.1.0.
arigo
- [pypy-commit] pypy default: Import cffi 1.1.0
arigo
- [pypy-commit] jitviewer default: One missing wrap_html()
arigo
- [pypy-commit] pypy stmgc-c8: Failed to fix the issue, so increase the trace_limit for now...
arigo
- [pypy-commit] pypy use_min_scalar: Start branch to correctly resolve the output dtype of ufunc(array, scalar) calls.
rlamy
- [pypy-commit] pypy use_min_scalar: Move scalar special-casing inside find_specialization()
rlamy
- [pypy-commit] pypy stmgc-c8: Found a way to, maybe, reduce the frequency of heapcache resets
arigo
- [pypy-commit] pypy stmgc-c8: Revert 77df620af990: it turns out that d2d1782d3d34 is mostly enough
arigo
- [pypy-commit] cffi release-1.1: Release branch
arigo
- [pypy-commit] cffi release-1.1: md5/sha of cffi-1.1.0.tar.gz
arigo
- [pypy-commit] cffi default: hg merge release-1.1
arigo
- [pypy-commit] pypy default: Use Py3k-compatible syntax here.
mjacob
- [pypy-commit] pypy py3k: Use Py3k-compatible syntax here.
mjacob
- [pypy-commit] pypy default: fix tests, error messages to be strictly numpy 1.9.0
mattip
- [pypy-commit] cffi default: english
arigo
- [pypy-commit] cffi default: Add a test that is hopefully passing on all platforms
arigo
- [pypy-commit] pypy default: Fix test_some_integer_type on 32-bit machines. Add a more precise test.
arigo
- [pypy-commit] cffi default: Leave with Py_FatalError() if we detect that the object is really
arigo
- [pypy-commit] pypy default: Skip instead of fail if the _*_cffi.so modules have not been made
arigo
- [pypy-commit] cffi default: Remove old 'flags=' which is just ignored by distutils
arigo
- [pypy-commit] pypy optresult: start working on raw virtuals
fijal
- [pypy-commit] cffi default: don't use A0, B0, CC0, D0 because termios.h might be included and it has
arigo
- [pypy-commit] pypy default: Moved tag release-2.6.0 to changeset 295ee98b6928 (from changeset e03971291f3a)
mattip
- [pypy-commit] buildbot default: typo
mattip
- [pypy-commit] pypy release-2.6.x: merge default into release (for documenting release)
mattip
- [pypy-commit] pypy release-2.6.x: document branch in release-2.6.0, not head
mattip
- [pypy-commit] cffi default: Trying to change the test to match the real use case more closely
arigo
- [pypy-commit] cffi default: Actually, we can test directly that getptr() returns the address of the
arigo
- [pypy-commit] pypy py3k: Add a failing test for a stack depth computation bug.
mjacob
- [pypy-commit] pypy py3k: (arigo, mjacob) Disable the StackDepthComputationError for now on this branch.
mjacob
- [pypy-commit] pypy py3k: Add a test for another stack depth computation error.
mjacob
- [pypy-commit] pypy py3k: Fix two previously added tests by making the stack depth computation more conservative (but still slightly incorrect).
mjacob
- [pypy-commit] pypy py3k: Remove test that doesn't apply to py3k.
mjacob
- [pypy-commit] pypy py3k: 2to3
mjacob
- [pypy-commit] pypy py3k: Found what looks like the proper fix. It is actually important to keep
arigo
Last message date:
Sun May 31 22:30:44 CEST 2015
Archived on: Sun May 31 22:31:41 CEST 2015
This archive was generated by
Pipermail 0.09 (Mailman edition).