An introduction to libuv (2016)
nikhilm.github.io
nikhilm.github.io
libuv v1.30 was release at on Jan 28, 2015. It is at v1.22.0.
A lot of the book was copied into the official docs later https://github.com/libuv/libuv/pull/1246.
The book definitely does not capture new things. I do not intend to update it, but I still look at pull requests.
How does it handle the well known incompatibility between fork and threads?
Forking for something else, though, god help you.
It looks like libuv creates a pipe for every new process, and uses that to send errno back on any failing system call. This has the disadvantage that you only get the error code, and none of the context. You get ENOENT, but you don't know which path was invalid.
(Or, if you want to be weird, do a PTRACE_TRACEME after fork and have the parent trace the child process and only detach when it sees a successful return from execve. If ptrace is unavailable, fall back to less-useful errors)
It's compatible with the native asyncio event loop and can be used as a drop-in replacement.
https://blogs.msdn.microsoft.com/webdev/2018/04/12/asp-net-c...
Much of the value in making a struct (and hence its size) public and supplying a pointer to the initializer is the potential to embed or avoid heap allocation of the thing altogether, and here in an example which would benefit from both the simplified code in addition to demonstrating the advantage, it's completely missed.
Fixed form:
int main() {
uv_loop_t loop;
uv_loop_init(&loop);
printf("Now quitting.\n");
uv_run(&loop, UV_RUN_DEFAULT);
uv_loop_close(&loop);
return 0;
}
mallocing is the entirely sensible thing to do here.
struct embbed { ... uv_loop_t loop; ... } e1, e2; ... e1 = e2; ?
As-is the API and the way its use is being demonstrated in that example appear amateur to say the least.
It's just a different programming approach in C.
Encapsulation is good only for making it harder for anyone to poke in the internals of your library, which has some benefits with binary compatibility, etc.
But it's strictly worse in all other metrics when coding in C. It limits you, it forces malloc calls where most of the time none would be necessary, it makes memory management more complicated, it forces explict initialization, etc.
What makes sense for a monolithic kernel's code is quite different from what makes sense in a userspace library, particularly on its public API boundaries.
It's also worth noting that the Linux kernel developers have historically made choices which are actively hostile to kernel module API/ABI stability as a way to discourage third parties from distributing out-of-tree proprietary binary modules. There is literally a policy of not having a stable interface for module writers, it's the anti-library except for the system call interface.
I've only worked with libuv in a few contexts, and maybe there are compelling reasons for exposing the struct. (I'd like to hear them! I would not have expected a uv_loop to be allocated in perf-critical code paths or code paths where failure isn't an option.) But I think critical analysis of C API design is an important topic. C gets a bad rap for being unsafe, which it obviously is in many ways, but as C developers, simple choices like this (that a novice might not even think much about) can make an API much safer -- or much less safe.
Struct embedding helps with this quite a bit. And it's not possible without exposing the struct definition.
Performance gains are possible, but that's secondary.
> If one already accepts that the object should be dynamically allocated
It can still be allocated on heap, but in one continuous chunk of memory as a part of the larger struct. Hiding the struct definition would prevent this.
That's largely true, but it's often cited as a reason to avoiding malloc/free (however dubious that is).
> To me the less you juggle with pointers and malloc/free the safer your code will be from memory leaks, misuse of unallocated memory, NULL checking issues, and easier to inspect/reason about.
I don't quite agree. I've worked mostly in code bases using the pattern described earlier (an opaque pointer, a $type_create(), and a $type_destroy() function). With that pattern, I find it much easier to be certain by code inspection that a particular object or transformation is valid because as long as the pointer was allocated correctly, the object can only be modified by functions that know the type (aside from memory corruption, but that's always possible). That's usually a small set of functions that know the struct details. This fact is useful both as a library author and as the author of a library consumer. By contrast, if the struct is exposed, it's harder to identify all the places that can modify the structure's details and to be sure that invariants are maintained in all those places.
Besides that, several other failure modes are much less likely with opaque structures, including copying a structure you shouldn't, miscopying a structure that's okay to copy (e.g., off-by-one while copying), or operating on a correctly-sized block of memory that's never been initialized. You can still use unallocated memory, of course, but that's fairly easy to make safer by initializing pointers to NULL. (The analogous option for stack-initialized structs -- initializing them to zero -- is not necessarily any safer than leaving them uninitialized -- particularly if the struct contains file descriptors.)
There are tradeoffs to both approaches. To me, the ability to modify the struct in new versions of the library without breaking the ABI is a pretty major point in favor of using an opaque structure for a library's primary handle. For very simple ancillary structures that are very unlikely to change, and where the convenience of stack allocation is worthwhile, exposing the structure makes a lot of sense.
The option to dynamically alloc uv_loop_t is not to avoid "accidental copies" or anysuch thing. Is there for when you need to dynamically create loop contexts.
Linux does exactly the same thing. But you wont see unecesary kmallocs of structs when static storage is enough. In such cases you will see the *_init functions used.