| Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
separate out `incremental_gc_step()` from `mrb_incremental_gc()`, and misc minor rename
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
rename advance_phace() to incremental_gc_until(), and cleanup some redundant codes with it.
|
|
|
|
|
|
in fact advance_phace() is not really a "phase", but can be used
as an utility function.
|
|
|
|
|
|
|
|
Refactor mrb_realloc() func in gc.c.
|
|
|
|
|
|
remove the redundant codes with mrb_malloc_simple()
|
|
|
|
|
|
calloc in C99 is expected to return NULL when nelem * len == 0.
|
|
|
|
|
|
|
|
|
|
|
|
instead of the struct.
|
|
not the full RVALUE known by the GC
|
|
|
|
|
|
freed. Amended the count_objects test to ensure the correct distinction
|
|
objects for each type
|
|
|
|
|
|
|
|
|
|
|
|
|
|
At the first time, mrb_realloc raise RuntimeError.
At the time fails twice in a row, it causes panic.
|
|
seems SIZE_MAX is defined in stdint.h. And it possibly (depends on the version of VC++) conflicts with SIZE_MAX in limits.h. This patch is no need if I did not support VC++.
|
|
|
|
|
|
|
|
|
|
|