Commits

Show all
Author Commit Message Labels Comments Date
Christian Tismer
rstacklet.py builds ok, but crashes when run. Checking in, anyway. Can somebody have a look? (transplanted from f1598797a877e03a21f4e9569268d4374ff75f36)
Branches
release-1.7.x
Christian Tismer
relevant_environ updated for windows (transplanted from d4d7276882d44a876d99da11875559967bab2e64)
Branches
release-1.7.x
Christian Tismer
got continuelets ready for windows. There were two things where I stumbled quite a while. 1) _compile_c_file was sloppy with the defined order of arguments, where masm had different behavior. 2) it was needed to add external definitions, but it was not clear to me that this already works and nothing than an entry in export_symbols was needed. (transplanted from b64cba156148599f77c3e1b6c5f7d5764609463d)
Branches
release-1.7.x
Christian Tismer
made the assembler sources compile so far (transplanted from 7d9e78a91dceea0c3b8d10cb5888317459f8f72c)
Branches
release-1.7.x
Maciej Fijalkowski
Added tag release-1.7 for changeset ff4af8f31882
Branches
release-1.7.x
Maciej Fijalkowski
fix the test
Tags
release-1.7
Branches
release-1.7.x
Maciej Fijalkowski
merge in default
Branches
release-1.7.x
Maciej Fijalkowski
Merge jitdriver-setparam-all branch that allows setting parameters to all jitdrivers.
Maciej Fijalkowski
close about to be merged branch
Branches
jitdriver-setparam-all
Armin Rigo
Re-add an assert that was lost.
Branches
jitdriver-setparam-all
Maciej Fijalkowski
rename numpy -> numpypy
Maciej Fijalkowski
Change set_param API. Now we run set_param(driver-or-None, 'name', value) instead of driver.set_param(...). This makes it possible to pass None, which changes the param on all drivers.
Branches
jitdriver-setparam-all
Amaury Forgeot d'Arc
issue887: cpyext: add support for mp_ass_subscript
Armin Rigo
Fix.
Armin Rigo
Workaround. See comment.
Armin Rigo
Fix syslog.py by following the CPython source, as per https://bugs.pypy.org/issue928. Not really tested -- there seems to be no tests for syslog?
Armin Rigo
Tweaks. Unsure why but it seems that test_zll_random ends up with two differently-typed arrays at the same address --- it might be because of casts. The _subarray caching fails in this case, by confusing the types of the subarrays.
Armin Rigo
Argh. See comment.
Armin Rigo
Fix for test_zrpy_releasegil.
Armin Rigo
Fix.
Armin Rigo
The hack "y + VERY_LARGE_FLOAT == y" fails to give the correct result on gcc, not only on msvc. Revert to the comparison with INFINITY and -INFINITY when not jitted.
Armin Rigo
Added a failing test.
Amaury Forgeot d'Arc
Update PyMODINIT_FUNC for C++ extensions.
Amaury Forgeot d'Arc
Allows presetup.py to execute setup.py scripts which use the "if __name__ == '__main__'" idiom
Alex Gaynor
Merged jit-dynamic-getarrayitem. Added support for creating custom getarrayitems at jit-compile time. Steals some stuff from anto's ffistruct.
Alex Gaynor
close for merge
Branches
jit-dynamic-getarrayitem
Alex Gaynor
put tests in a sane subclass and only run the new ones in the x86 backend
Branches
jit-dynamic-getarrayitem
Alex Gaynor
merged default in
Branches
jit-dynamic-getarrayitem
Alex Gaynor
handle hashing a None rstr, fixes the test I checked in yesterday.
Alex Gaynor
simplify failing test
  1. Prev
  2. Next