Commits

Show all
Author Commit Message Labels Comments Date
Hakan Ardo
hg merge jit-short_from_state
Branches
jit-str_in_preamble
Hakan Ardo
hg merge default
Branches
jit-short_from_state
Hakan Ardo
rpythonized
Branches
jit-short_from_state
Hakan Ardo
moved VirtualState handling into file of its own separating it (more or less) from the resume handling
Branches
jit-short_from_state
Hakan Ardo
produce first part of the jumpargs from the VirtualState
Branches
jit-short_from_state
Armin Rigo
merge heads
Armin Rigo
Fix for failing tests.
Alex Gaynor
math.is{inf,nan} are purefunctions
Alex Gaynor
Don't generate nonsense for UNPACK_ITERABLE with a non-{list,tuple} item (specifically array)
Hakan Ardo
move inputargs generation to VirtualState to make it reusable to create the jumpargs
Branches
jit-short_from_state
Hakan Ardo
hg merge default
Branches
jit-str_in_preamble
Hakan Ardo
forcing unsupported vritauls changes the VirtualState
Branches
jit-str_in_preamble
Hakan Ardo
proper cloning of values when transferin them to next iteration
Branches
jit-str_in_preamble
Hakan Ardo
testing boxed virtual strings
Branches
jit-str_in_preamble
Armin Rigo
Allow this file to be imported if 'float' has no '__getformat__', as is the case in pypy 1.4.1.
Armin Rigo
Trying to fix issue677: remove the hard-coded endianness from dtoa.c, and instead generate a stub .c file that defines the correct endianness and #includes dtoa.c.
Armin Rigo
Move this hack into an official interface.
Armin Rigo
Test and implementation for "i0 > i0" and friends. Not implemented for unsigned comparisons so far (does not seem really useful).
Hakan Ardo
hg merge default
Branches
jit-str_in_preamble
Hakan Ardo
typos
Branches
jit-str_in_preamble
Armin Rigo
The purpose of these tests is not to count the number of operations.
Armin Rigo
Backout c8e4be12f7b1, whose purpose is actually to test that we get the correct results.
Armin Rigo
Fix the test that broke when changing how "x/y" is implemented. Rewrite it in a way that doesn't use "/" at all.
Armin Rigo
Kill the test for division_to_rshift, because it's already tested directly in test_optimizeopt.py. Also, it seems that the trace only contained additions, because the actual divisions were all constant-folded, so the test was not testing much.
Armin Rigo
On x86-64, put the correct dfi assembler macros to allow gdb to debug the function (and walk past it in the backtrace).
Armin Rigo
Fix running py.test in this directory.
Hakan Ardo
workaround allowing the unrolling to fail and produce a preamble calling itself instead of chrasing
Branches
jit-str_in_preamble
Hakan Ardo
another problematic case
Branches
jit-str_in_preamble
Hakan Ardo
this case wount work
Branches
jit-str_in_preamble
Hakan Ardo
additional passing test
Branches
jit-str_in_preamble
  1. Prev
  2. Next