mirichi
ad60f81032
The correct is to compare the MRB_STACK_GROWTH.
2014-04-26 19:37:53 +09:00
Yukihiro "Matz" Matsumoto
e76be4e2ab
check length before calling stack_copy(); based on a patch from @mirich; close #2129
2014-04-26 12:07:51 +09:00
mirichi
fa1a1e6db8
Refactoring OP_CALL
2014-04-25 21:49:29 +09:00
Yukihiro "Matz" Matsumoto
83c1399af0
eliminate plain int except for a few cases like arena_index
2014-04-25 04:01:57 +09:00
mirichi
4ab801c045
fixed.
2014-04-23 19:39:14 +09:00
mirichi
e6c757b5c1
bug fixed
2014-04-22 23:10:40 +09:00
mirichi
8195224072
reduce the init size of stack
2014-04-22 22:02:03 +09:00
mirichi
b523018803
OP_SEND refactoring
2014-04-22 20:04:10 +09:00
cremno
bc23a5e9e5
remove mrb_str_buf_cat
...
It does the same as `mrb_str_cat`.
2014-04-21 21:28:06 +02:00
mirichi
b796c503fa
OP_EQ系が速くなった
2014-04-15 19:30:42 +09:00
Takeshi Watanabe
824d3450c5
Qualify argv argument of API const.
2014-04-11 20:38:23 +09:00
take_cheeze
9f24cd6d79
Qualify mrb_yield_*'s argv const.
2014-04-10 17:34:38 +09:00
Yukihiro "Matz" Matsumoto
2d61698799
should not access proc->body.irep if proc is PROC_CFUNC; close #1999
2014-04-04 01:47:47 +09:00
take_cheeze
755ded2ef0
Use MRB_ENV_STACK_SHARED_P to check cioff field of REnv.
2014-04-01 01:53:28 +09:00
take_cheeze
b38eb72d82
Add macro MRB_ENV_UNSHARE_STACK instead of assigning -1 directly.
2014-04-01 01:35:40 +09:00
take_cheeze
80534b6d14
Use MRB_ENV_STACK_LEN instead of accessing flags directly to get REnv's stack length.
2014-04-01 01:13:39 +09:00
Yukihiro "Matz" Matsumoto
49d18ca848
Kernel#send should preserve nregs in callinfo; close #1970
2014-03-29 00:21:52 +09:00
Yukihiro "Matz" Matsumoto
b8e434a5e6
initialize ci->target_class in mrb_toplevel_run; ref #1942 close mattn/mruby-require#20
2014-03-27 23:51:08 +09:00
Yukihiro "Matz" Matsumoto
28871b9d1f
should not over process ensure clauses; close #1949
2014-03-26 17:49:27 +09:00
Yukihiro "Matz" Matsumoto
bfd2a539ed
add new function mrb_toplevel_run to prevent running through C function boudaries on exceptions; close #1942
2014-03-26 01:09:09 +09:00
take_cheeze
f536492521
use FiberError in vm.c
2014-03-22 03:19:58 +09:00
Yukihiro "Matz" Matsumoto
36836c8eb9
should restore proc after rescue; close #1891
2014-03-20 16:10:12 +09:00
Yukihiro "Matz" Matsumoto
9084e03556
rename mrb_yield_internal to mrb_yield_with_class since it's no longer internal
2014-03-19 08:57:54 +09:00
cubicdaiya
019d15c3ee
Use mrb_exc_new_str_lit widely
2014-03-04 02:45:53 +09:00
Yukihiro "Matz" Matsumoto
369c6c3649
Kernel#send does not use mrb_funcall anymore for most cases; ref #1680
2014-03-03 13:31:29 +09:00
take_cheeze
b98890a4b7
clear exc_catched before jumping to L_RAISE
2014-03-01 20:53:31 +09:00
take_cheeze
78915f9601
support c++ exception
2014-03-01 20:05:29 +09:00
Yukihiro "Matz" Matsumoto
675fd3dc80
allow send method not to call mrb_funcall if calling method is implemented in Ruby; fix #1680 ref #1765
2014-03-01 18:13:10 +09:00
Yukihiro "Matz" Matsumoto
8e170ce915
exception in fiber should make resume to raise in parent fiber context; close #1765
2014-03-01 12:58:31 +09:00
cubicdaiya
67de10bfcb
use C style comments instead of C++ style comments
...
According to CONTRIBUTING.md,
Don't use C++ style comments
/* This is the prefered comment style */
Use C++ style comments only for temporary comment e.g. commenting out some code lines.
2014-03-01 03:31:45 +09:00
Yukihiro "Matz" Matsumoto
6bd3d88edd
support break from fiber block; fix #1766
2014-02-28 18:43:35 +09:00
Yukihiro "Matz" Matsumoto
03fde35760
restore proc after restoring from fiber; ref #1766
2014-02-28 12:50:32 +09:00
Yukihiro "Matz" Matsumoto
d264e1addd
fix indent
2014-02-27 03:50:59 +09:00
cubicdaiya
4c283714f6
use mrb_str_new_lit instead of mrb_str_new for C string literal
2014-02-26 22:34:12 +09:00
take_cheeze
08ea324593
move src/error.h to include/mruby/error.h
2014-02-20 22:41:29 +09:00
Yukihiro "Matz" Matsumoto
e29efb02bd
normalize NaN after division that may generate NaN; fix #1712
2014-02-15 19:59:26 +09:00
cremno
8a3a3a5a49
Clang and ICC/ICL: define DIRECT_THREADED
...
This is most likely only needed on Windows be-
cause __GNUC__ is not defined by both compilers.
It might fail with some unusual configurations.
If that is the case, feel free to open an issue.
2014-02-11 23:55:52 +01:00
Yukihiro "Matz" Matsumoto
00bdaec766
initialize ci->stackent at the top; #1691
2014-02-09 19:20:16 +09:00
Miura Hideki
5df82e357c
Store raw stack address in callinfo instead of offset from stbas
2014-02-09 16:42:25 +09:00
Yukihiro "Matz" Matsumoto
04eb577b69
Fiber.new{break} caused SEGV
2014-02-05 12:51:17 +09:00
Yukihiro "Matz" Matsumoto
e0e5aebc66
add callback invocation from OP_DEBUG
2014-01-23 15:08:38 +09:00
cremno
aa655b9ee1
remove superfluous includes
...
- reduce compile time by a little bit (full-core: ~0.7s for me)
- thanks to 'include-what-you-use' for some help
- include Standard C header files before any other (coding style)
2014-01-07 17:56:30 +01:00
Yukihiro "Matz" Matsumoto
791021575c
need not to initialize ci->err at toplevel
2014-01-05 20:03:45 +09:00
Yukihiro "Matz" Matsumoto
f5cca078e3
avoid decrimenting eidx out of ecall()
2014-01-05 19:18:13 +09:00
h2so5
0e23038a60
execute ensure clause correctly without OP_EPOP
2014-01-05 16:24:20 +09:00
Yukihiro "Matz" Matsumoto
2122cf48db
pop ensure stack before execution; close #1638
2014-01-04 01:30:38 +09:00
Yukihiro "Matz" Matsumoto
6bcbef5064
always clear ci->err
2014-01-04 00:54:05 +09:00
Yukihiro "Matz" Matsumoto
4b68fbcd83
quicker release of arena after ensure calls
2014-01-04 00:13:28 +09:00
h2so5
77f215c837
use macros for assignments to ci->err
2014-01-03 21:54:58 +09:00
nanami
87a392067a
Fixed callinfo buffer overflow while calling ensure handlers
2013-12-15 22:25:59 +08:00