It's safer to check the buffer size is large enough to hold the error
message before creating the slice from raw parts.
Also, this patch remove the need for `last_error`, simplifying the
code a little bit. The `length` variable is casted to `usize` once.
`wasmer-runtime` has a `debug` and a `llvm` features. Let's mirror
them in `wasmer-runtime-c-api` so that the user can, for instance,
compile with the LLVM backend.
This function is required to transform a `wasmer_byte_array` into a
`wasmer_serialized_module_t`. This is the complementary function of
`wasmer_serialized_module_bytes`.
The `wasmer_module_serialize` function now computes a
`wasmer_serialized_module_t` value. The `wasmer_module_deserialize`
function takes this value as an input. Same for
`wasmer_serialized_module_destroy`.
The new function `wasmer_serialized_module_bytes` allows to read the
bytes inside the `wasmer_serialized_mdule_t` structure.
This test suite compiles a module, then serializes it, deserializes
it, and continues by creating an instance and calling a function on
it. It allows to test the entire roundtrip.
This patch updates the first argument of `wasmer_validate` from `*mut
uint8_t` to `*const uint8_t`. Indeed, the
`wasmer-runtime-core::validate` function doesn't expect a mutable
slice, so it's not required to expect a mutable array from C.
Also, it's likely for the Wasm bytes to be stored in the
`wasmer_byte_array` structure. The first field `bytes` is defined as
`*const uint8_t`. So this patch avoids a cast when writing a C++
program.
This patch adds 2 functions for the runtime C API, respectively
`wasmer_instance_context_data_set` and
`wasmer_instance_context_data_get`. The goal is to modify the
`vm::Ctx.data` field in the `runtime-core` library.
This is required to pass dynamic data to imported functions for
instance.
`imports` contains `func_import`, `global_import`, `memory_import` and
`table_import`, so 4 items. This patch updates the length of `imports`
when calling the `wasmer_instantiate` function.