At a fundamental level, runtime-known stack allocation harms code reusability.
Edit: commenters identified 2 more puzzle pieces below, but there's still one that didn't get asked about yet :P
@alloca(T: type, count: usize, upper_bound_count: comptime_int)
when I had been only thinking about zig for 2 years, I thought the same.
> It's too tempting to use incorrectly.
A compile-time-determined upper bound would solve this.
> The stack is allocated based on a compile-time-determined upper bound.
A compile-time-determined upper bound would solve this too.
Shouldn't a performance-oriented language give the programmer tools to improve memory locality? And what's wrong with spexguy's idea?
there are no good choices in the case where you really need that thing you claim to need. recognizing that fact and picking different strategy is good engineering.
> what are you going to do with the rest of the stack?
I'll leave it for the rest of the system. My app will use less memory, and since memory locality is improved, there will be fewer cache misses, meaning it runs faster too.
> let's say you take a function call that is about to overflow the stack
Stack overflows are impossible thanks to the comptime upper_bound parameter. That's the entire premise of this thread.
I thought Zig was all about maximum performance. Sometimes I just want a little bit of stack memory, which will often already be in L1 cache.
Sigh. So I have to choose between something I think might be useful, for something that too many languages have already soiled themselves with. Hopes that Zig has a better solution, but not optimistic.
Our stack compels me to work in Swift, Kotlin, Elixir, and Python. I use the async feature of Swift and Kotlin when some library forces me to. I actually preferred working with GCD before Swift had to join the async crowd. Elixir of course just has this problem solved already.
I frequently ask others who work in these languages how often they themselves reach for the async abilities of their languages, and the best I ever get from the more adventurous type is “I did a play thing to experiment with what I could do with it”.
C libraries?
Even on languages without VLAs one can implement a simulacra of them with recursion.
So it will be the same thing but with more (error handling) steps.
This annoyance can be avoided by avoiding recursion. Where recursion is useful, it can be done, you just have to handle failure properly, and then you'll have safety against stack overflow.
> Where recursion is useful, [...]
Recursion is so useful, most imperative languages even have special syntax constructs very specific special cases of recursion they call 'loops'.
Yes[1]. You can use the @call builtin with the .always_tail modifier.
@call(.always_tail, foo, { arg1, arg2, ... });
How do incremental compilation and distributed compilation work?
That’s a genuinely interesting point. I don’t think known sizes for locals are a hard requirement here, though threading this needle in a lower-level fashion than Swift would need some subtle language design.
Fundamentally, what you want to do is construct an (inevitably) runtime-sized type (the coroutine) out of (by problem statement) runtime-sized pieces (the activation frames, itself composed out of individual, possibly runtime-sized locals). It’s true that you can’t then allow the activations to perform arbitrary allocas. You can, however, allow them to do allocas whose sizes (and alignments) are known at the time the coroutine is constructed, with some bookkeeping burden morally equivalent to maintaining a frame pointer, which seems fair. (In Swift terms, you can construct a generic type if you know what type arguments are passed to it.) And that’s enough to have a local of type of unknown size pulled in from a dynamic library, for example.
Again, I’m not sure how a language could express this constraint on allocas without being Swift (and hiding the whole thing from the user completely) or C (and forcing the user to maintain the frames by hand), so thank you for drawing my attention to this question. But I’m not ready to give up on it just yet.
> At a fundamental level, runtime-known stack allocation harms code reusability.
This is an assertion, not an argument, so it doesn’t really have any points I could respond to. I guess my view is this: there are programs that can be written with alloca and can’t be written without (unless you introduce a fully general allocator, which brings fragmentation problems, or a parallel stack, which is silly but was in fact used to implement alloca historically). One other example I can give in addition to locals of dynamically-linked types is a bytecode interpreter that allocates virtual frames on the host stack. So I guess that’s the other side of being opinionated—those whose opinions don’t match are turned away.
Frankly, I don’t even know why I’m defending alloca this hard. I’m not actually happy with the status quo of just yoloing a hopefully maybe sufficiently large stack. I guess the sticking point is that you seem to think alloca is obviously the wrong thing, when it’s not even close to obvious to me what the right thing is.
(Once upon a time, MSLU, a Microsoft-provided Unicode compatibility layer for Windows 9x, used stack-allocated buffers to convert strings from WTF-16 to the current 8-bit encoding. That was also a bad idea.)