i like the way zig interfaces with c. even though it’s a “newer” (not 1.0) language you can import native c which means you have access to already really fast libraries. the build script and macros are the same language you write your code. the best part is you can interface with normal code (not unsafe) and can create boundaries where sure in the c side you can’t enforce memory safety, but where you interface you can convert into memory safe types.
to talk more about memory safety i think rust does a great job, however there are some inconsistentcies where a lot of memory allocations can throw out of memory errors that will crash the app whereas zig you could gracefully handle it. this is probably more important for lower level things like kernel than an app where not recovering means everything goes down.
honestly i wish there was an opt in borrow checker as some patterns are easily expressed, but honestly i would opt out for async code as it just gets verbosely over-declarative imo. also the new async in zig is really really cool.
rust does a great job, however there are some inconsistentcies where a lot of memory allocations can throw out of memory errors that will crash the app
That is not an “inconsistency”. Crashing the application when you run out of memory is honestly in almost all cases the correct call and crashing is memory safe. Most applications can do absolutely nothing when running out of memory. “Handling” OOM is extremely complicated and most applications simply cannot handle it in any way.
Of course low level stuff needs to sometimes actually handle this, but it’s mostly an operating system thing. And Rust still allows you that control, it’s just not the default behavior of the collections and such in the standard library, because again, it almost never makes sense to try to handle OOM. But if you really need that behavior, you can have it.
You can trivially use C libraries in Rust, or any system language that supports the C ABI for that matter, and this includes many hobbyist languages. Zig is not special.
No Zig definitely is special. You don’t have to do any extra busy work to call C code - you can pretty much just #include the header and that’s that. It’s similar to calling C from C++.
In any other languages - including Rust - you have to do some work declaring functions, wrapping them and so on. It’s not hard but it definitely is a non-zero amount of tedious work (especially before AI). There’s absolutely no way you could describe it as “trivial”, unless someone else has already done that work for you.
But I don’t think it is a significant Zig advantage really. When I’m writing Rust, it’s extremely rare that I want to call any C code that someone else hasn’t already done the tedious wrapping for.
zig is a bit special in this regard though as it is a c compiler not just using ffi. this means you could swap out gcc, make autotools etc and drop in zig with a build.zig. mostly useful if you want to migrate large code bases as you can incrementally port the parts you want.
if you’re picking it up for a new project then yeah this probably doesn’t impact you much.
Beyond what amounts to a packaging difference, what does that even mean?
How do you think zero-cost C ABI support (including fully working cross-lang LTO) works in lang implementations that have that? And how do you think that’s different from what Zig gives you?
Note: “Zero-cost” means zero runtime cost, not zero compile cost. The compiler does real work — monomorphization and inlining — to make abstractions disappear, and that work shows up as longer build times. See Reducing Compile Time for the trade-off.
most of the benefits come from a shared code base, one compiler (which is insanely fast with incremental rebuilds and can watch files for changes) means changes in the c code reflect in a single build step within seconds. the benefit is primarily for the developer as the compiler output should be relatively similar.
I’m not sure what you’re talking about then, in zigs case you don’t need an exposed ABI to use existing c code?
in terms of performance it will always be ambiguous as implementations are hard to compare across languages, however they use llvm as an alternative to support architectures as they build out their native compiler. in the transition for x86_64 they were seeing up to 70% increase in compile speeds. its still ambiguous as its dependent on their implementation of llvm, though its such a large increase it is still meaningful.
but i say all this as a rust and zig advocate where i would happily choose either over the the majority of alternatives.
it will always be ambiguous as implementations are hard to compare across languages
Correct.
in the transition for x86_64 they were seeing up to 70% increase in compile speeds
And that was a part of what I was hinting at, because you get >>70% speed-up with Cranelift in most Rust projects. But in either case, faster code generation is not free lunch, hence the mention of comparable runtime performance of generated binaries.
i like the way zig interfaces with c. even though it’s a “newer” (not 1.0) language you can import native c which means you have access to already really fast libraries. the build script and macros are the same language you write your code. the best part is you can interface with normal code (not unsafe) and can create boundaries where sure in the c side you can’t enforce memory safety, but where you interface you can convert into memory safe types.
to talk more about memory safety i think rust does a great job, however there are some inconsistentcies where a lot of memory allocations can throw out of memory errors that will crash the app whereas zig you could gracefully handle it. this is probably more important for lower level things like kernel than an app where not recovering means everything goes down.
honestly i wish there was an opt in borrow checker as some patterns are easily expressed, but honestly i would opt out for async code as it just gets verbosely over-declarative imo. also the new async in zig is really really cool.
That is not an “inconsistency”. Crashing the application when you run out of memory is honestly in almost all cases the correct call and crashing is memory safe. Most applications can do absolutely nothing when running out of memory. “Handling” OOM is extremely complicated and most applications simply cannot handle it in any way.
Of course low level stuff needs to sometimes actually handle this, but it’s mostly an operating system thing. And Rust still allows you that control, it’s just not the default behavior of the collections and such in the standard library, because again, it almost never makes sense to try to handle OOM. But if you really need that behavior, you can have it.
You can trivially use C libraries in Rust, or any system language that supports the C ABI for that matter, and this includes many hobbyist languages. Zig is not special.
No Zig definitely is special. You don’t have to do any extra busy work to call C code - you can pretty much just
#includethe header and that’s that. It’s similar to calling C from C++.In any other languages - including Rust - you have to do some work declaring functions, wrapping them and so on. It’s not hard but it definitely is a non-zero amount of tedious work (especially before AI). There’s absolutely no way you could describe it as “trivial”, unless someone else has already done that work for you.
But I don’t think it is a significant Zig advantage really. When I’m writing Rust, it’s extremely rare that I want to call any C code that someone else hasn’t already done the tedious wrapping for.
You know
bindgenexists, right?Yes. It’s not as easy and reliable as Zig.
zig is a bit special in this regard though as it is a c compiler not just using ffi. this means you could swap out gcc, make autotools etc and drop in zig with a build.zig. mostly useful if you want to migrate large code bases as you can incrementally port the parts you want.
if you’re picking it up for a new project then yeah this probably doesn’t impact you much.
Beyond what amounts to a packaging difference, what does that even mean?
How do you think zero-cost C ABI support (including fully working cross-lang LTO) works in lang implementations that have that? And how do you think that’s different from what Zig gives you?
from rust docs:
most of the benefits come from a shared code base, one compiler (which is insanely fast with incremental rebuilds and can watch files for changes) means changes in the c code reflect in a single build step within seconds. the benefit is primarily for the developer as the compiler output should be relatively similar.
That has nothing to do with what we were talking about. And in any case:
Needs qualification vs. other languages for binaries with comparable runtime performance.
(Hint: you will be surprised.)
Not special or unique.
Not only not special, but literally exists in all workable build tools.
bindgen+build.rsis the Rust version of this.I’m not sure what you’re talking about then, in zigs case you don’t need an exposed ABI to use existing c code?
in terms of performance it will always be ambiguous as implementations are hard to compare across languages, however they use llvm as an alternative to support architectures as they build out their native compiler. in the transition for x86_64 they were seeing up to 70% increase in compile speeds. its still ambiguous as its dependent on their implementation of llvm, though its such a large increase it is still meaningful.
but i say all this as a rust and zig advocate where i would happily choose either over the the majority of alternatives.
ABI is not something that gets “exposed” or not.
Correct.
And that was a part of what I was hinting at, because you get >>70% speed-up with Cranelift in most Rust projects. But in either case, faster code generation is not free lunch, hence the mention of comparable runtime performance of generated binaries.