summaryrefslogtreecommitdiffhomepage
path: root/src
AgeCommit message (Collapse)Author
2021-12-20vm.c: fix a half-baked implementation of `OP_SETIDX`; ref #5608Yukihiro "Matz" Matsumoto
2021-12-20codedump.c: adjust the position of local variable labels.Yukihiro "Matz" Matsumoto
2021-12-15Merge pull request #5599 from dearblue/loss-digitsYukihiro "Matz" Matsumoto
Avoid losing the upper digits for mruby binary
2021-12-14proc.c: add `mrb_state` argument to `mrb_proc_copy()`.Yukihiro "Matz" Matsumoto
The function may invoke the garbage collection and it requires `mrb_state` to run.
2021-12-13Avoid losing the upper digits for mruby binarydearblue
- `rlen` keeps 16 bits. - `ilen` keeps 32 bits. Note that this change will break mruby binary format compatibility.
2021-12-13variable.c: resurrect `size` member in `iv_tbl`.Yukihiro "Matz" Matsumoto
The existence of this member reduces memory and execution time.
2021-12-13class.c: increase first allocated page size.Yukihiro "Matz" Matsumoto
2021-12-13class.c: implement method cache (off by default).Yukihiro "Matz" Matsumoto
2021-12-08vm.c (check_method_noarg): the value at `kidx` may not be a hash.Yukihiro "Matz" Matsumoto
2021-12-08vm.c: use `check_method_noarg()` to reduce code duplication; ref #5584Yukihiro "Matz" Matsumoto
2021-12-08vm.c: fix `mrb_ci_kidx`.Yukihiro "Matz" Matsumoto
It used to return wrong value for 14 positional arguments.
2021-12-08variable.c: fix clang integer warning.Yukihiro "Matz" Matsumoto
2021-12-07variable.c: avoid redundant iv scan in `mrb_mod_cv_set()`.Yukihiro "Matz" Matsumoto
Now `iv_get()` returns `pos+1` if it finds the entry, so you don't need to call `iv_put()`. You can replace the entry value by assigning to `t->ptr[pos-1]`.
2021-12-04variable.c: reduce array access in iv hash table.Yukihiro "Matz" Matsumoto
2021-12-03variable.c: avoid `mrb_undef_value()` for delete entries.Yukihiro "Matz" Matsumoto
Instead embed deleted flag in the key (`mrb_sym` only occupies 30bits).
2021-12-03variable.c: remove `size` member from `iv_tbl` to reduce memory.Yukihiro "Matz" Matsumoto
`iv_size()` is approximated by the allocated table size.
2021-12-03variable.c: first iv allocation size should be 4 instead of 1.Yukihiro "Matz" Matsumoto
2021-12-03variable.c: should not access `NULL[0]` to avoid asan warnings.Yukihiro "Matz" Matsumoto
2021-12-01codedump.c: print `OP_EXT` explicitly again for debugging purpose.Yukihiro "Matz" Matsumoto
2021-12-01variable.c: reduce memory usage of instance variable tableYukihiro "Matz" Matsumoto
This is a fundamentally simplified reimplementation of #5317 by @shuujii Instead of having array of `struct iv_elem`, we have sequences of keys and values packed in single chunk of malloc'ed memory. We don't have to worry about gaps from alignment, especially on 64 bit architecture, where `sizeof(struct iv_elem)` probably consumes 16 bytes, but `sizeof(mrb_sym)+sizeof(mrb_value)` is 12 bytes. In addition, this change could improve memory access locality. close #5317
2021-11-29vm.c: remove duplicate code in VM.Yukihiro "Matz" Matsumoto
2021-11-28Align "wrong number of arguments" messagesdearblue
Make "N for M" into the form "given N, expected M". As I worked, I noticed that the `argnum_error()` function had a part to include the method name in the message. I think this part is no longer needed by https://github.com/mruby/mruby/pull/5394. - Before this patch ```console % bin/mruby -e '[1, 2, 3].each 0' trace (most recent call last): [1] -e:1 -e:1:in each: 'each': wrong number of arguments (1 for 0) (ArgumentError) ``` - After this patch ```console % bin/mruby -e '[1, 2, 3].each 0' trace (most recent call last): [1] -e:1 -e:1:in each: wrong number of arguments (given 1, expected 0) (ArgumentError) ```
2021-11-26Check more `MRB_ARGS_NONE()`dearblue
The `__id__` method implemented in the C function has `MRB_ARGS_NONE()` specified, but it is also effective in the following cases. ```ruby p nil.__id__ opts: 1 rescue p :a p nil.method(:__id__).call 1 rescue p :b p nil.method(:__id__).call opts: 1 rescue p :c p nil.method(:__id__).to_proc.call 1 rescue p :d p nil.method(:__id__).to_proc.call opts: 1 rescue p :e p nil.method(:__id__).unbind.bind_call nil, 1 rescue p :f p nil.method(:__id__).unbind.bind_call nil, opts: 1 rescue p :g p nil.__send__ :__id__, 1 rescue p :h p nil.__send__ :__id__, opts: 1 rescue p :i ``` After applying this patch, all items will output symbols in the same way as CRuby. For this purpose, add `MRB_PROC_NOARG` to `struct RProc::flags`.
2021-11-20Merge pull request #5578 from dearblue/OP_SENDBYukihiro "Matz" Matsumoto
Fixes keywords are lost with the `OP_SENDB`
2021-11-19Fixed a discrepancy in `OP_ASET`dearblue
There was a discrepancy in the actual behavior, assertions, and documentation. Therefore, I modified it based on the actual behavior.
2021-11-19Fixes keywords are lost with the `OP_SENDB`dearblue
If a splat argument was passed, it could write out of range on the VM stack. ```console % bin/mruby -e 'def m(*args, **opts, &blk) p [args, opts, blk] end; m(*%w(X Y Z), r: 1, g: 2, b: 3) {}' [["X", "Y", "Z"], {}, #<Proc:0x80077d7d0>] ```
2021-11-09etc.c: improve `mrb_obj_id` with NaN boxing and Word boxing.Yukihiro "Matz" Matsumoto
2021-11-04hash.c: avoid `mrb_obj_id` to get the hash value if possible.Yukihiro "Matz" Matsumoto
2021-11-03Merge pull request #5565 from dearblue/class.newYukihiro "Matz" Matsumoto
Improved `Class#new` method
2021-11-01object.c: fast implementation of `mrb_obj_eq`.Yukihiro "Matz" Matsumoto
For `MRB_NAN_BOXING` and `MRB_WORD_BOXING`.
2021-10-30boxing_nan.h: allow `MRB_INT64` with `MRB_NAN_BOXING`.Yukihiro "Matz" Matsumoto
Integers out of 32 bit range will be allocated in the heap.
2021-10-29Improved `Class#new` methoddearblue
The number of registers used is reduced. Also, previously `R6` and` R7` were used, which exceeded the limit of `new_irep.nregs = 6`. This could cause the VM stack to overrun.
2021-10-24Make `mrb_static_assert()` a variable argumentdearblue
`mrb_static_assert()` extends the macro function to take one or two arguments. If the argument is other than that, an error will occur. References: - static_assert のメッセージ省略を許可 - cpprefjp C++日本語リファレンス https://cpprefjp.github.io/lang/cpp17/extending_static_assert.html - c - Overloading Macro on Number of Arguments - Stack Overflow https://stackoverflow.com/a/11763277
2021-10-23string.c: use FNV1a algorithm for the string hash function.Yukihiro "Matz" Matsumoto
2021-10-21codedump.c: add indentation for `OP_MOVE`.Yukihiro "Matz" Matsumoto
2021-10-19codedump.c: remove `OP_` prefix from disasm output.Yukihiro "Matz" Matsumoto
2021-10-19ops.h: add new instructions `OP_SSEND` and `OP_SSENDB`.Yukihiro "Matz" Matsumoto
These instructions call methods of the receiver.
2021-10-14vm.c: need to ensure the stack size is sufficient.Yukihiro "Matz" Matsumoto
Need to add `n` because the stack will be shifted in the `mrb_funcall`.
2021-10-12error.c: small refactoring regarding #2485Yukihiro "Matz" Matsumoto
2021-10-12Merge branch 'exc_mesg' of https://github.com/take-cheeze/mruby into ↵Yukihiro "Matz" Matsumoto
take-cheeze-exc_mesg
2021-10-12Support Ruby3.0 keyword arguments.Yukihiro "Matz" Matsumoto
The Difference Since Ruby1.9, the keyword arguments were emulated by Ruby using the hash object at the bottom of the arguments. But we have gradually moved toward keyword arguments separated from normal (positinal) arguments. At the same time, we value compatibility, so that Ruby3.0 keyword arguments are somewhat compromise. Basically, keyword arguments are separated from positional arguments, except when the method does not take any formal keyword arguments, given keyword arguments (packed in the hash object) are considered as the last argument. And we also allow non symbol keys in the keyword arguments. In that case, those keys are just passed in the `**` hash (or raise `ArgumentError` for unknown keys). The Instruction Changes We have changed `OP_SEND` instruction. `OP_SEND` instruction used to take 3 operands, the register, the symbol, the number of (positional) arguments. The meaning of the third operand has been changed. It is now considered as `n|(nk<<4)`, where `n` is the number of positional arguments, and `nk` is the number of keyword arguments, both occupies 4 bits in the operand. The number `15` in both `n` and `nk` means variable sized arguments are packed in the object. Positional arguments will be packed in the array, and keyword arguments will be packed in the hash object. That means arguments more than 14 values are always packed in the object. Arguments information for other instructions (`OP_SENDB` and `OP_SUPER`) are also changed. It works as the third operand of `OP_SEND`. the difference between `OP_SEND` and `OP_SENDB` is just trivial. It assigns `nil` to the block hidden arguments (right after arguments). The instruction `OP_SENDV` and `OP_SENDVB` are removed. Those instructions are replaced by `OP_SEND` and `OP_SENDB` respectively with the `15` (variable sized) argument information. Calling Convention When calling a method, the stack elements shall be in the order of the receiver of the method, positional arguments, keyword arguments and the block argument. If the number of positional or keyword arugument (`n` or `nk`) is zero, corresponding arguments will be empty. So when `n=0` and `nk=0` the stack layout (from bottom to top) will be: +-----------------------+ | recv | block (or nil) | +-----------------------+ The last elements `block` should be explicitly filled before `OP_SEND` or assigned to `nil` by `OP_SENDB` internally. In other words, the following have exactly same behavior: OP_SENDB clears `block` implicitly: ``` OP_SENDB reg sym 0 ``` OP_SEND clears `block` implicitly: ``` OP_LOADNIL R2 OP_SEND R2 sym 0 ``` When calling a method with only positional arguments (n=0..14) without keyword arguments, the stack layout will be like following: +--------------------------------------------+ | recv | arg1 | ... | arg_n | block (or nil) | +--------------------------------------------+ When calling a method with arguments packed in the array (n=15) which means argument splat (*) is used in the actual arguments, or more than 14 arguments are passed the stack layout will be like following: +-------------------------------+ | recv | array | block (or nil) | +-------------------------------+ The number of the actual arguments is determined by the length of the argument array. When keyword arguments are given (nk>0), keyword arguments are passed between positional arguments and the block argument. For example, when we pass one positional argument `1` and one keyword argument `a: 2`, the stack layout will be like: +------------------------------------+ | recv | 1 | :a | 2 | block (or nil) | +------------------------------------+ Note that keyword arguments consume `2*nk` elements in the stack when `nk=0..14` (unpacked). When calling a method with keyword arguments packed in the hash object (nk=15) which means keyword argument splat (**) is used or more than 14 keyword arguments in the actual arguments, the stack layout will be like: +------------------------------+ | recv | hash | block (or nil) | +------------------------------+ Note for mruby/c When mruby/c authors try to support new keyword arguments, they need to handle the new meaning of the argument information operand. If they choose not to support keyword arguments in mruby/c, it just raise error when `nk` (taken by `(c>>4)&0xf`) is not zero. And combine `OP_SENDV` behavior with `OP_SEND` when `n` is `15`. If they want to support keyword arguments seriously, contact me at <[email protected]> or `@yukihiro_matz`. I can help you.
2021-10-10vm.c: unify `mrb_gc_arena_restore` and `mrb_gc_arena_shrink`.Yukihiro "Matz" Matsumoto
2021-10-09Merge pull request #5555 from dearblue/str-ref-regexpYukihiro "Matz" Matsumoto
Partially allow overriding of `String#[]` methods
2021-10-09Fix typo for `OP_SETIDX`dearblue
2021-10-09Partially allow overriding of `String#[]` methodsdearblue
This is for the purpose of supporting `Regexp`. - configuration and build ```console % cat regexp_config.rb MRuby::Lockfile.disable MRuby::Build.new do toolchain "clang" enable_debug gem core: "mruby-bin-mruby" gem core: "mruby-print" gem mgem: "mruby-onig-regexp" #gem mgem: "mruby-regexp-pcre" end % rake MRUBY_CONFIG=regexp_config.rb ``` - mruby HEAD (bec074e) ```console % build/host/bin/mruby -e 'p "abcdefg"[/.(?=...$)/]' -e:1: can't convert OnigRegexp into Integer (TypeError) ``` - with this patch ```console % build/host/bin/mruby -e 'p "abcdefg"[/.(?=...$)/]' "d" ```
2021-10-08vm.c: check call stack depth before pushing the frame.Yukihiro "Matz" Matsumoto
Existing call stack depth checks are unified into this check in `cipush()`. The maximum depth is now specified by `MRB_CALL_LEVEL_MAX` (the default is 512). The older `MRB_FUNCALL_DEPTH_MAX` is no longer used.
2021-10-06codedump.c: remove trailing extra newline.Yukihiro "Matz" Matsumoto
The bug was introduced in 8be78bd.
2021-10-03mruby/ops.h: add new instructions `OP_GETIDX` and `OP_SETIDX`.Yukihiro "Matz" Matsumoto
Which represent `obj[int]` and `obj[int]=val` respectively where `obj` is either `string`, `array` or `hash`, so that index access could be faster. When `obj` is not assumed type or `R(a+1)` is not integer, the instructions fallback to method calls.
2021-09-30value.h: introduce `MRB_TT_STRUCT` for `mruby-struct`.Yukihiro "Matz" Matsumoto
2021-09-27parse.y: `__ENCODING__` should be determined in run-time; fix #5552Yukihiro "Matz" Matsumoto
It used to be compiled to the static string in the compiler. But the encoding status actually depends on the runtime configuration. A new method `Kernel#__ENCODING__` is introduced to implement the feature.