~ensoft-opensource/+junk/cpython-boxes : changes

~ensoft-opensource/+junk/cpython-boxes » Changes from revision 51

From Revision 51 to 32

Rev	Summary	Authors	Date
51	Adding partial support for 64-bit. Adding partial support for 64-bit.	Peyman Owladi	10 years ago
50	Fix shortcircuiting in long_pow (fixes test_pow) Fix shortcircuiting in long_pow (fixes test_pow)	Phil Connell	10 years ago
49	Skip test_logging testcase failing in baseline Skip test_logging testcase failing in baseline	Phil Connell	10 years ago
48	Mark some places needing changes for test_gdb Mark some places needing changes for test_gdb No attempt at finding everything/fixing the issues yet :)	Phil Connell	10 years ago
47	Fix for pow, given a negative mantissa and modulus Fix for pow, given a negative mantissa and modulus e.g. pow(-10, 0, -7) should give -6, not 1.	Phil Connell	10 years ago
46	Fix for test_fractions -- long_true_div must return floats. Fix for test_fractions -- long_true_div must return floats.	Phil Connell	10 years ago
45	Fix for test_decimal (overflow in addition) Fix for test_decimal (overflow in addition) Problem was: >>> 16000000 + 777216 0 Uhh... Added a (debug) check for overflow in Py_UNBOXED_CREATE for good measure.	Phil Connell	10 years ago
44	Hack in fix for test_xml_etree_c Hack in fix for test_xml_etree_c The _elementtree extension module does its own playing with the low bit of object pointers. So, need to make sure that we box appropriately. There's a follow-on cleanup patch needed: should have a convenience function/macro to box an unboxed value of any type.	Phil Connell	10 years ago
43	Fix stupid regression from SIZE changes Fix stupid regression from SIZE changes	Phil Connell	10 years ago
42	Fix (partially) Py_SIZE to respect unboxed objects. Fix (partially) Py_SIZE to respect unboxed objects. i. Updated Py_SIZE for longobject.c only (since lvalue uses need changing and cba) ii. Added Py_SIZE_UPDATE to replace Py_SIZE lvalues. iii. Use Py_SIZE_UPDATE where appropriate in longobject.c	Phil Connell	10 years ago
41	Stupid hacky solution for an unboxed ->ob_digit ref. Stupid hacky solution for an unboxed ->ob_digit ref. Need something better.	Phil Connell	10 years ago
40	Fix some raw ob_type references (causing segfaults) Fix some raw ob_type references (causing segfaults)	Phil Connell	10 years ago
39	Disable randomly crashing test (doesn't seem related to boxi... Disable randomly crashing test (doesn't seem related to boxing, maybe upstream)	Phil Connell	10 years ago
38	Deal with possibility that __int__ can return an unboxed lon... Deal with possibility that __int__ can return an unboxed long. Problem was that our shortcircuit was too early: the passed-in object wasn't an unboxed long, but its __int__ returned one. Correct thing was to delay the short-circuit check until after __int__ is called. WARNING: HERE BE DRAGONS. There are probably further instances of this bug. Crash was in test_getargs2, so maybe we should add a wider range of similar cases (given that test_long didn't catch the problem...)	Phil Connell	10 years ago
37	Skip test_long -> test_small_ints Skip test_long -> test_small_ints Problem is that there are now two 'instances' of each small int: - The cached, boxed version - The unboxed version Since not all operators (e.g. //) return unboxed ints, can have e.g. 0 is not 0 // 1	Phil Connell	10 years ago
36	Fix off-by-one error for powers of two in msb() Fix off-by-one error for powers of two in msb()	Phil Connell	10 years ago
35	More ob_type dereferences; fixes for decimal More ob_type dereferences; fixes for decimal	Martin Morrison	10 years ago
34	More test fixes: return modifiable Unicodes from __str__; ch... More test fixes: return modifiable Unicodes from __str__; change some ob_refcnt raw accesses	Martin Morrison	10 years ago
33	A bunch of edge-cases fixes. Now passes the first 8 tests in... A bunch of edge-cases fixes. Now passes the first 8 tests in the test suite.	Martin Morrison	11 years ago
32	Fix long_pow and a few other things Fix long_pow and a few other things	Martin Morrison	11 years ago

Older »