513c639bf9
patch for last bug corrected to be against last released version (and not against last version in RCS...)
Roberto Ierusalimschy
2015-04-13 13:13:25 -03:00
69b5f7a410
some details in 'luaO_int2fb' + more consistent use of the locale decimal point
Roberto Ierusalimschy
2015-04-11 15:30:08 -03:00
ae76c39712
Bug: suspended '__le' metamethod can give wrong result
Roberto Ierusalimschy
2015-04-10 14:56:25 -03:00
cfc84c856d
'LUA_REAL_*' -> 'LUA_FLOAT_*' (everywhere else we are using 'float')
Roberto Ierusalimschy
2015-03-31 09:00:07 -03:00
63720a4290
janitor work (comments, variable names, some other details)
Roberto Ierusalimschy
2015-03-30 16:51:00 -03:00
484bf14a6b
calls to 'luaC_checkGC' in luaD_precall moved near to 'luaD_checkstack' (which is what can need memory)
Roberto Ierusalimschy
2015-03-30 13:05:23 -03:00
abb2f5511d
detail ('b + base' -> 'base + b' to follow all other similar uses)
Roberto Ierusalimschy
2015-03-30 12:45:01 -03:00
fcc6e1220e
avoid expression with side effect (in debug mode) inside 'sizeof'
Roberto Ierusalimschy
2015-03-30 12:43:51 -03:00
b436ed58a3
'clearapihash' -> 'luaS_clearcache' and moved to 'lstring.c' (which keeps all code related to this cache)
Roberto Ierusalimschy
2015-03-25 10:42:19 -03:00
57c0db219b
line history keep lines without added 'return'
Roberto Ierusalimschy
2015-03-09 18:57:05 -03:00
4ba0cb4580
always use macros to operate on lua_Numbers
Roberto Ierusalimschy
2015-03-07 16:30:16 -03:00
a30c66f0fc
macro 'luai_apicheck'/'api_check' back with a 'lua_State' parameter (some people use it)
Roberto Ierusalimschy
2015-03-06 16:49:50 -03:00
bb4baa73ea
allocation function is not exactly API (and cannot raise errors like other API functions); better not use 'api_check' for cheking it.
Roberto Ierusalimschy
2015-03-06 16:45:54 -03:00
d7d44b038d
detail (use 'lua_pushliteral' to push an empty string)
Roberto Ierusalimschy
2015-03-06 16:09:08 -03:00
297512b34c
avoid using 'lua_pushliteral' in a non-zero-terminated string
Roberto Ierusalimschy
2015-03-05 13:07:46 -03:00
2ae2e6408e
avoid testing for NULL when marking objects that cannot be NULL
Roberto Ierusalimschy
2015-03-04 10:51:55 -03:00
a80cada914
new cache for interning strings
Roberto Ierusalimschy
2015-03-04 10:31:21 -03:00
a00013c8d0
'point2int' -> 'point2uint' (to reflect its actual result type)
Roberto Ierusalimschy
2015-03-03 16:53:13 -03:00
b9a1f27250
do not attempt emergency collection while building state (it is useless, and state can be inconsistent)
Roberto Ierusalimschy
2015-03-03 15:18:29 -03:00
951b562cf8
configuration for numerical types through only one #define for each type
Roberto Ierusalimschy
2015-03-02 13:59:01 -03:00
6408bc0b7f
new macros 'chgfltvalue'/'chgivalue' (numerical for loop does not need to set the type of its internal variable at each iteration)
Roberto Ierusalimschy
2015-03-02 13:04:52 -03:00
91efb4b895
LUA_COMPAT_5_1 needs some options deprecated from 5.2 to 5.3
Roberto Ierusalimschy
2015-02-28 16:22:31 -03:00
27c5b1b237
added patches to two bugs
Roberto Ierusalimschy
2015-02-23 16:20:43 -03:00
d51bdc166d
bug: 'string.format("%f")' can cause a buffer overflow (with long doubles) bug: 'debug.getlocal' on a coroutine suspended in a hook can crash the interpreter
Roberto Ierusalimschy
2015-02-20 16:21:29 -02:00
81245b1ad5
'numisinteger' (for table keys) replaced by 'luaV_tointeger' (old 'tointeger_aux'), which can do the same job.
Roberto Ierusalimschy
2015-02-20 12:27:53 -02:00
397ce11996
make 'hashfloat' configurable
Roberto Ierusalimschy
2015-02-20 12:05:01 -02:00
97f2aa5a44
bug: when manipulating other threads, there is no garanties about their stack space
Roberto Ierusalimschy
2015-02-19 15:06:21 -02:00
419e2cb01d
detail (file does not need 'string.h')
Roberto Ierusalimschy
2015-02-16 11:15:00 -02:00
4bdf9962bf
bug: suspended function can have its 'func' field not pointing to its function, crashing debug functions
Roberto Ierusalimschy
2015-02-13 14:01:17 -02:00
e0306e386f
some changes in 'hashfloat' to avoid undefined (in ISO C) numerical conversions
Roberto Ierusalimschy
2015-02-13 11:05:34 -02:00
2e6e53c7cc
added API checks to some unprotected 'top' increments
Roberto Ierusalimschy
2015-02-11 16:47:22 -02:00
b7dfd18612
make buffer for format specification a little larger (length modifier can be larger than 2, e.g. "I64")
Roberto Ierusalimschy
2015-02-04 10:54:31 -02:00
8efcd411fe
own implementation for 'string.format("%a")' for C89 platforms
Roberto Ierusalimschy
2015-02-04 10:52:57 -02:00
90d016980f
detail ('luaL_getmetatable' does not return a boolean, but a tag)
Roberto Ierusalimschy
2015-02-03 15:38:24 -02:00
b3996c55f6
LUAL_BUFFERSIZE affects the API, so it is better not to change it between releases of the same version. (The long double case seems to be too rare, so that should not be a problem; moreover, that case is related to a bug)
Roberto Ierusalimschy
2015-01-16 15:26:56 -02:00
438a9fcc78
LUAI_MAXSHORTLEN moved from 'luaconf.h' to 'llimits.h' (does not affect the API and is seldom modified)
Roberto Ierusalimschy
2015-01-16 15:15:52 -02:00
7e2015a46d
size of short strings stored in a single byte, to reduce the size of struct 'TString'
Roberto Ierusalimschy
2015-01-16 14:54:37 -02:00
5b01cb39b5
larger MINSTRTABSIZE (standard Lua starts with ~230 strings, after oppening the standard libraries)
Roberto Ierusalimschy
2015-01-16 13:41:03 -02:00
965f74525b
removed macro LUAI_FIRSTPSEUDOIDX (that stuff does not need to be configurable)
Roberto Ierusalimschy
2015-01-13 17:50:30 -02:00
3f2e21f0ea
Long double needs a larger LUAL_BUFFERSIZE (because 'string.format("%.99f", 1e4900)' can generate quite long strings), but otherwise buffer can be 1~2K
Roberto Ierusalimschy
2015-01-13 15:24:22 -02:00
3509914916
BUG (when compiled with long double): buffer overflow when formatting string.format("%.99f", 1e4930)
Roberto Ierusalimschy
2015-01-13 15:18:25 -02:00
ae27be40c9
better check for overflows in 'table.move' (removes restriction that initial position should be positive)
Roberto Ierusalimschy
2015-01-13 14:27:29 -02:00
cdd26700e8
definition for 'MAXUPVAL' moved for a more "private" place and its value and comment corrected to reflect current implementation
Roberto Ierusalimschy
2015-01-13 13:49:11 -02:00
a9a4bf68b0
typo in comment
Roberto Ierusalimschy
2015-01-13 11:18:04 -02:00
f31cc8e342
typo in error message
Roberto Ierusalimschy
2015-01-12 17:32:32 -02:00
0adfa773b9
new macro 'cast_func' adds '__extension__' (in gnu compilers) when converting void* to function, to avoid warnings
Roberto Ierusalimschy
2015-01-05 11:51:39 -02:00
3317f5c6d9
includes 'stddef.h' (as it uses NULL)
Roberto Ierusalimschy
2015-01-05 11:48:33 -02:00
dbf911c516
detail (added -Wconversion as a comment, to be used ocasionally)
Roberto Ierusalimschy
2015-01-02 11:03:52 -02:00
3a29087cb7
'lua_setlocal' should not pop value when failing (to be consistent with the manual and with 'lua_setupvalue')
Roberto Ierusalimschy
2015-01-02 10:52:22 -02:00
234fb7f695
clearer(?) code (also avoids a warning about 'c' being used without initialization)
Roberto Ierusalimschy
2015-01-02 10:50:28 -02:00
c077d47465
does not define _XOPEN_SOURCE when LUA_USE_C89 is defined + defining _XOPEN_SOURCE==0 undefines that macro
Roberto Ierusalimschy
2014-12-29 14:54:13 -02:00
8e5290d81e
all "divisions" (div,idiv,mod) by zero are not folded, to avoid problems during compilation + does not fold zero results, as they can collapse with -0.0 and the ANSI test to distinguish both needs a division by zero (which we are trying to avoid) + removed macro 'luai_numinvalidop' (as its main use case were divisions by zero)
Roberto Ierusalimschy
2014-12-29 14:49:25 -02:00
a1c37f834a
detail (macro 'luai_numidiv' uses 'luai_numdiv' to compute the division)
Roberto Ierusalimschy
2014-12-29 11:27:55 -02:00
1a7868c1d5
bug: 'random' limit is 2^31-1, not RAND_MAX
Roberto Ierusalimschy
2014-12-27 18:32:26 -02:00
50e2eb8954
change in macro 'vmcase', avoding code inside it. (Code inside the macro harms tools such as debuggers.)
Roberto Ierusalimschy
2014-12-27 18:30:38 -02:00
94eb37dc73
new year (2015)
Roberto Ierusalimschy
2014-12-26 15:24:27 -02:00
a51069202a
removed macro 'luai_numinvalidop' (main motivation removed, as folding does not handle any division by zero by default)
Roberto Ierusalimschy
2014-12-26 12:44:44 -02:00
88ef06f4f3
comments were wrong (not updated about several changes)
Roberto Ierusalimschy
2014-12-20 11:58:15 -02:00
eb45f3a9b6
details in 'luaM_reallocvchar'
Roberto Ierusalimschy
2014-12-19 15:26:14 -02:00
6321041058
new macro 'luaM_reallocvchar' to allocate arrays of chars (avoids uneeded tests and respective warnings)
Roberto Ierusalimschy
2014-12-19 11:45:40 -02:00
2b83711fba
new macro 'nvalue' (to convert an object to a float when we know object is a number)
Roberto Ierusalimschy
2014-12-19 11:36:32 -02:00
3ae21a352c
detail (in test for 'luai_numinvalidop', use a round float)
Roberto Ierusalimschy
2014-12-19 11:33:06 -02:00
1465edf341
old Visual Studio versions did not support 'noreturn' attribute
Roberto Ierusalimschy
2014-12-19 11:30:23 -02:00
77e786d436
stupid bug in T.stacklevel (not in use by the tests)
Roberto Ierusalimschy
2014-12-18 10:13:42 -02:00
6d79752e65
Windows uses 'long long' (instead of '__int64') when present; again, uses macro LLONG_MAX as a proxy for the support for 'long long'.
Roberto Ierusalimschy
2014-12-16 15:17:30 -02:00
75812a517b
added macro LUA_USE_I64 for separate control of use of Windows-specific types
Roberto Ierusalimschy
2014-12-16 14:20:01 -02:00
151dc5cc5f
traverse loaded modules (instead of globals) for a name for a function + removes prefix '_G.' from names (if present)
Roberto Ierusalimschy
2014-12-14 16:32:26 -02:00
Roberto Ierusalimschy