| Age | Commit message (Collapse) | Author |
|---|
|
|
|
- mrb_utf8len() - returns the size of a UTF-8 char (in bytes)
- mrb_utf8_strlen() - returns the length of a UTF-8 string (in char)
|
|
|
|
Because we can't use the highest 2 bits of function pointers.
|
|
|
|
that includes `float.h`. It allows definitions from native headers.
|
|
|
|
The basic idea of this change is from @dearblue.
Note: the arguments of `mrb_str_pool()` have changed, but the function
is provided for internal use (No `MRB_API`). So basically you don't have
to worry about the change.
|
|
|
|
shuujii/add-mrb_num_args_error-for-wrong-number-of-arguments-error
Add `mrb_num_args_error()` for "wrong number of arguments" error
|
|
#### Example (with `MRB_METHOD_CACHE`)
```ruby
GC.start
c = Class.new
p c #=> #<Class:0x7fd6a180e790>
c.new #=> cache `c.new`
c = nil
GC.start #=> `c` is GCed
r = Range.dup
p r #=> #<Class:0x7fd6a180e790>
# [same pointer as `c`]
r.new(2, 3) #=> ArgumentError: 'initialize':
# wrong number of arguments (2 for 0)
# [`c.new` is called instead of `r.new`]
```
#### Cause
An entry of method cache is identified by class pointer and method
id. However, reusing memory after GC may create a class with the same
pointer as the cached class.
#### Treatment
Cleared method caches of the class when the class is GCed.
|
|
Because location info (file name and line number) is kept in the backtrace,
it should not be kept in the result of `inspect` (and the exception object
itself), I think.
### Example
```ruby
# example.rb
begin
raise "err"
rescue => e
p e
end
```
#### Before this patch:
```
$ bin/mruby example.rb
example.rb:2: err (RuntimeError)
```
#### After this patch:
```
$ bin/mruby example.rb
err (RuntimeError)
```
|
|
To unify the style of messages.
|
|
|
|
### Example
```ruby
# example.rb
max32 = 2**30 - 1 + 2**30
min32 = -max32-1
[max32, max32+1, min32, min32-1].each{|n| p [n, n.class]}
```
#### Before this patch:
```
$ bin/mruby example.rb
[2147483647, Float]
[2147483648, Float]
[-2147483648, Float]
[-2147483649, Float]
```
#### After this patch:
```
$ bin/mruby example.rb
[2147483647, Fixnum]
[2147483648, Float]
[-2147483648, Fixnum]
[-2147483649, Float]
```
|
|
The bit width terminology is unified to `BIT` according to `MRB_INT_BIT`
and `CHAR_BIT`. Also the bit position terminology is unified to `BIT_POS`.
|
|
shuujii/introduce-mrb_ssize-type-for-buffer-size-on-memory
Introduce `mrb_ssize` type for buffer size on memory; ref #4483
|
|
Previously, `mrb_int` was used as the type that represents the buffer size
on memory, but the sizes of `RString` and `RArray` exceed 6 words when
`MRB_INT64` is enabled on 32-bit CPU.
I don't think it is necessary to be able to represent the buffer size on
memory that exceeds the virtual address space. Therefore, for this purpose,
introduce `mrb_ssize` which doesn't exceed the sizes of `mrb_int` and
pointer.
I think all `mrb_int` used for this purpose should be changed to
`mrb_ssize`, but currently only the members of the structures (`RString`,
`mrb_shared_string`, `RArray` and `mrb_shared_array`) are changed.
|
|
shuujii/always-enable-the-rational-and-complex-literals
Always enable the rational and complex literals
|
|
I think they can always be enabled because the regular expression literal is
always enabled.
|
|
|
|
|
|
|
|
|
|
Keyword arguments can now be retrieved with the `:` specifier and
`mrb_kwargs` data.
For the interface, I referred to CRuby's `rb_get_kwargs()`.
For implementation, I referred to `OP_KARG` or etc.
|
|
This is an experimental changes in Ruby 2.7.
|
|
|
|
|
|
|
|
The type `mrb_hash_value` is no longer used by the segmented list
implementation (ref e8dcfe1 and e65d426).
|
|
The return value from `mrb_delete_key` needs to be protected from GC in
some cases.
|
|
shuujii/use-type-predicate-macros-instead-of-mrb_type-if-possible
Use type predicate macros instead of `mrb_type` if possible
|
|
|
|
For efficiency with `MRB_WORD_BOXING` (implement type predicate macros for
all `enum mrb_vtype`).
|
|
Use inttypes for `snprintf()`
|
|
Remove `MRB_TT_HAS_BASIC` macro
|
|
* mrb_sym2name -> mrb_sym_name
* mrb_sym2name_len -> mrb_sym_name_len
* mrb_sym2str -> mrb_sym_str
|
|
|
|
The value of `MRB_TT_HAS_BASIC` is meaningless because `MRB_TT_HAS_BASIC` is
no longer used with `MRB_WORD_BOXING` at b2c3d88f.
|
|
The default implementations of type predicate macros use `mrb_type`.
But `mrb_type` with `MRB_WORD_BOXING` isn't very efficient, so the new
implementations avoid `mrb_type`.
|
|
linked programs
In lld linked programs, .rodata comes before .text, thus mrb_ro_data_p
will return false for strings in .rodata. Change the lower bound from
_etext to __ehdr_start to catch these cases. This works for ld.bfd, gold
and lld, and it does not have false positives even if .init_array does
not exist.
Remove the branch that uses _edata: strings in .data can be modified so
this is semantically incorrect. Delete the __APPLE__ branch (its
manpages say get_etext() and get_edata() are strongly discouraged).
.init_array has been adopted by most ELF platforms to supersede .ctors.
Neither _etext nor _edata is used, so rename MRB_USE_ETEXT_EDATA to
MRB_USE_EHDR_START.
|
|
`MRB_METHOD_TABLE_INLINE` was fragile. It requires `-falign-functions=n`.
On platform that uses higher bits of function pointers, you can use new
`MRB_METHOD_T_STRUCT` configuration macro.
|
|
This is partial `aspec` check that only checks `MRB_ARGS_NONE()`.
|
|
|
|
So you don't need `-falign-functions=2` anymore. Instead your platform
must not use higher bits of the pointer (true for most platforms).
If not, you have to use `struct mrb_method_t` version.
|
|
|
|
|
|
|
|
|
|
|
