Commit Graph

4280 Commits

Author SHA1 Message Date
Yukihiro "Matz" Matsumoto 32aed6ad8c mruby-bigint: increase KARATSUBA_THREASHOLD to 512
Since Karatsuba algorithm is far heavier than the basic multiplication,
we have increased KARATSUBA_THREASHOLD and decreased MAX_RECURSION_DEPTH.
2024-10-04 18:10:14 +09:00
Yukihiro "Matz" Matsumoto 6820da1526 mruby-bigint (mpz_abs): use zero_p() 2024-10-04 18:10:14 +09:00
Yukihiro "Matz" Matsumoto fd8d3a8d70 mruby-bigint: use zero() to make mpz_t zero. 2024-10-04 18:10:14 +09:00
Yukihiro "Matz" Matsumoto 6274f8de43 mruby-bigint (mpz_sub): avoid data copy
Since mpz_sub() calls mpz_add() with opposite sign, refer same data from
the original, instead of copying whole data.
2024-10-04 18:10:14 +09:00
Yukihiro "Matz" Matsumoto 363754b5d1 mruby-bigint (uzero_p): check from the top, not bottom
For most of the case, the first non-zero value would be found earlier
when search starts from the top.
2024-10-04 18:10:14 +09:00
Yukihiro "Matz" Matsumoto fcc1832efc mruby-bigint (mul_karatsuba): avoid copying operand u and v 2024-09-30 18:32:26 +09:00
Yukihiro "Matz" Matsumoto 0023cbdb3d mruby-bigint (mpz_mul): use Karatsuba algorithm to improve performance 2024-09-28 05:26:40 +09:00
Yukihiro "Matz" Matsumoto 9c5812a463 mruby-compier: keep stack reference to passed block after modifying &b
To mark `MRB_PROC_ORPHAN` we need to keep track of passed block, even
after the assignment to the block argument. And `yield` should use the
original block; #5786, #5791, #6369
2024-09-26 02:48:16 +09:00
John Bampton e12da35645 Fix spelling 2024-09-24 00:14:24 +10:00
Yukihiro "Matz" Matsumoto f6caf0d788 mruby-bigint (mrb_init_set_str): reduce mpz_t allocations 2024-09-19 20:52:11 +09:00
Yukihiro "Matz" Matsumoto ba8ee7b895 mruby-socket (inet_pton): need to initialize loop variable 2024-09-17 12:05:46 +09:00
Yukihiro "Matz" Matsumoto afefb5261b mruby-rational (rat_sub_b): avoid normalization of big integers 2024-09-15 03:04:32 +09:00
Yukihiro "Matz" Matsumoto bdbf8536a7 mruby-bigint: rename add_d and sub_d to add_n and sub_n respectively
`_n` suffix represents `no_normalize`.
2024-09-15 03:02:41 +09:00
Yukihiro "Matz" Matsumoto 50225acec4 mruby-rational (rat_mul_b): should not normalize multiplication results 2024-09-15 02:50:57 +09:00
Yukihiro "Matz" Matsumoto 704ad87150 Merge pull request #6338 from dearblue/proc-align 2024-09-12 06:43:42 +09:00
Yukihiro "Matz" Matsumoto ee9b1381ec Merge pull request #6347 from leviongit/fixes/yield-kw-codegen
codegen.c,parse.y: remove flattening of `yield` arguments; fix #6346
2024-09-12 06:39:27 +09:00
Yukihiro "Matz" Matsumoto 413b005801 Merge pull request #6349 from dearblue/dir-test 2024-09-11 07:12:31 +09:00
Yukihiro "Matz" Matsumoto a38dd0c988 Merge pull request #6350 from dearblue/io-filepath 2024-09-11 07:10:45 +09:00
dearblue 9cb42438e1 Remove unnecessary mrb_gc_arena_restore()
When a function defined as a method returns, there is no need to call `mrb_gc_arena_restore()` immediately before it.
2024-09-10 21:45:44 +09:00
dearblue 33cf729a39 Fixed character encoding conversion function mismatch 2024-09-10 21:34:05 +09:00
dearblue 3aa9a5ee68 Perform GC before deleting directories
In some environments, the test will fail because the directory in use cannot be deleted.
This problem was encountered when building 32-bit binary with mingw32 on FreeBSD and running on wine.
2024-09-10 21:17:10 +09:00
Yukihiro "Matz" Matsumoto da4cfbf89c mruby-array-ext: add a new method Array#fetch_values 2024-09-10 10:22:34 +09:00
leviongit 87801ed5f1 codegen.c,parse.y: remove flattening of yield arguments; fix #6346
this commit changes the parse tree of the `yield` node, allowing for
proper keyword argument generation
2024-09-09 10:06:31 +02:00
dearblue e476d9a344 Need to restore the GC arena after some function calls
When calling `mrb_equal()` or `mrb_funcall()` family functions, the GC arena should be restored if the loop is repeated by a non-immediate return value.

In my opinion, restoring the GC arena is unnecessary when a non-immediate (true) value causes the function to return (e.g. the `mrb_ary_index_m()` function).

The patch does not take into account the case of recursive calls and may be incomplete.
2024-09-03 21:29:06 +09:00
Yukihiro "Matz" Matsumoto 1d12097bae mruby-io (mrb_file__gethome): allow avoiding getpwnam(3); ref #5358
You have to define `MRB_IO_NO_PWNAM` to skip calling getpwname(3).
2024-09-03 15:27:29 +09:00
Yukihiro "Matz" Matsumoto f2d9995984 Merge pull request #6332 from dearblue/mruby-compiler 2024-09-01 22:45:07 +09:00
dearblue 42513d61fd Need to place static proc objects into 8-byte alignments
Static proc objects defined as methods may be placed in 4-byte alignments in 32-bit environments.
This may be misinterpreted as an immediate value depending on the address.

Since C11 and C++11 have additional language features for byte alignment, corresponding compilers use them to define the `mrb_alignas()` macro.
For earlier compilers, they use their own extensions to define the `mrb_alignas()` macro.

GCC supports `__attribute__((aligned(alignment)))` since at least version 2.95.3 (1999).
https://gcc.gnu.org/onlinedocs/gcc-2.95.3/gcc_4.html#IDX305
According to GPT-4, support was added in version 2.7 (1995).

It is not known which version of Visual C++ added support for `__declspec(align(n))`.
According to GPT-4, at least Visual C++ 6.0 (1998) seems to support it.
Also, the documentation of past Intel C/C++ compilers that support `__declspec(align(n))` makes reference to support with Visual C++ 4.2 (1996).
https://www.intel.com/content/dam/www/public/ijkk/jp/ja/documents/developer/ccomp40j.pdf
2024-09-01 20:28:34 +09:00
Yukihiro "Matz" Matsumoto 4a71d4604e mruby-rational: add typecast to check if a value is fit in mrb_int 2024-08-27 22:48:40 +09:00
dearblue cfe410f466 Must not depend on the “host” build to generate mruby-compiler/core/y.tab.c
“mruby-compiler” should be able to generate `y.tab.c` files through a separate build configuration if it is not added to the ‘host’ build.
In the example above, the “host/mrbc” build should generate the `y.tab.c` file.
2024-08-25 11:17:56 +09:00
dearblue 79d7ac5f7e Restore the GC arena with tests.
This was the unintentional catalyst for finding the problem in #6329, but it is preferable to add a test if necessary.
2024-08-25 11:08:40 +09:00
Yukihiro "Matz" Matsumoto 204b81c563 mruby-bin-mirb: fix the issue on macOS; #6327 2024-08-24 07:13:45 +09:00
Marcelo Juchem a6eb3e5c27 fix ncurses linking issues
The build configuration for `mruby` assumes only the `ncurses` library
needs to be linked because `tinfo` is implicitly pulled in.

In environments where ncurses is available only as a static library,
`tinfo` needs to be linked explicitly (needed for functions like
`tputs`.

This patch fixes that by linking `tinfo` if available.

It also fixes the build for environments where only the `ncursesw`
version of the library (including wide character support) is present,
while still giving preference to the `ncurses` version (without wide
character support).
2024-08-22 14:11:48 -05:00
Yukihiro "Matz" Matsumoto 4ea6d74d83 mruby-rational: introduce the constants ONE and ZERO 2024-08-21 18:38:51 +09:00
Yukihiro "Matz" Matsumoto 39b9a1a06a mruby-rational (rational_new): fix compilation condition 2024-08-17 16:29:51 +09:00
Yukihiro "Matz" Matsumoto 44908e5cba mruby-rational: support bigint numerators & denominators 2024-08-16 15:12:03 +09:00
Yukihiro "Matz" Matsumoto 92f476d8fa mruby-bigint (mpz_set): trim destination mpz_t 2024-08-16 07:32:05 +09:00
Yukihiro "Matz" Matsumoto e9c4486e27 mruby-bigint: add fast path to mrb_bint_mul() & mrb_bint_div() 2024-08-14 22:36:47 +09:00
Yukihiro "Matz" Matsumoto 6e5a493ff3 mruby-bigint (mrb_bint_new_float): add fast path for FIXABLE_FLOAT 2024-08-14 16:11:27 +09:00
Yukihiro "Matz" Matsumoto 56a82009aa Revert "The lex_state after literals should be EXPR_ENDARG."; fix #6298
This reverts commit e98823f189.
2024-08-10 19:17:49 +09:00
Yukihiro "Matz" Matsumoto 6da2e60edc mruby-bigint (mpz_get_str): avoid malloc when base is a power of 2 2024-08-05 10:43:47 +09:00
Yukihiro "Matz" Matsumoto aa06432776 mruby-array-ext: index should be within mrb_int range
Type check is done by `__to_int` method.
2024-08-01 07:34:20 +09:00
Yukihiro "Matz" Matsumoto 3c5baac5c5 mruby-sprintf: fix a bug with negative octals
Leading octal digit may be 1, 3, 7 unlike binary or hexadecimal.
2024-07-30 11:33:18 +09:00
Yukihiro "Matz" Matsumoto ab95e042ca mruby-bigint (mrb_bint_xor): add fast path with fixnums 2024-07-29 16:44:54 +09:00
Yukihiro "Matz" Matsumoto b706c89c41 mruby-bigint (mrb_bint_2comp): fix assertion 2024-07-29 16:44:54 +09:00
Yukihiro "Matz" Matsumoto bf7f14341d mruby-bigint (mrb_bint_or): add fast path with fixnums 2024-07-29 15:47:19 +09:00
Yukihiro "Matz" Matsumoto dcf7adeeb7 mruby-bigint: use zero_p() 2024-07-29 15:43:22 +09:00
Yukihiro "Matz" Matsumoto 196c2bd0b9 mruby-bigint (mrb_bint_add): add type cast to stop warnings in C++ 2024-07-28 23:29:54 +09:00
Yukihiro "Matz" Matsumoto 081cd0627f mruby-bigint (mrb_bint_hash): need to get hash from all digits 2024-07-27 18:54:50 +09:00
Yukihiro "Matz" Matsumoto 203a89a3f2 mruby-bigint (zero_p): define a new predicate to check if mpz_t is 0 2024-07-27 18:54:50 +09:00
Yukihiro "Matz" Matsumoto 234a2b10fb mruby-bigint (mpz_neg): no need to check x != y 2024-07-27 18:54:50 +09:00