ccf670c made using `return` from within a comptime block in a non-inline
function illegal, since it is a use of runtime control flow in a
comptime block. It is allowed if the function in question is `inline`,
since no actual control flow occurs in this case. A few functions from
std (notably `std.fmt.comptimePrint`) needed to be marked `inline` to
support this change.
These functions are currently footgunny when working with pointers to
arrays and slices. They just return the stated length of the array/slice
without iterating and looking for the first sentinel, even if the
array/slice is a sentinel terminated type.
From looking at the quite small list of places in the standard
library/compiler that this change breaks existing code, the new code
looks to be more readable in all cases.
The usage of std.mem.span/len was totally unneeded in most of the cases
affected by this breaking change.
We could remove these functions entirely in favor of other existing
functions in std.mem such as std.mem.sliceTo(), but that would be a
somewhat nasty breaking change as std.mem.span() is very widely used for
converting sentinel terminated pointers to slices. It is however not at
all widely used for anything else.
Therefore I think it is better to break these few non-standard and
potentially incorrect usages of these functions now and at some later
time, if deemed worthwhile, finally remove these functions.
If we wait for at least a full release cycle so that everyone adapts to
this change first, updating for the removal could be a simple find and
replace without needing to worry about the semantics.
I originally started monkeying with this because std.mem.zeroes doesn't support sentinel-terminated const slices even with defaults in 0.10.x.
I see that std.mem.zeroes was modified in #13256 to allow setting these slices to "".
That got me partway to where I wanted, but there was still an issue fields whose types are structs, they wouldn't get their defaults.
So when iterating struct fields looking for default values, when there is no default value and the type is .Struct, it will delegate to a call to zeroInit.
* Initialize struct fields in zeroInit exactly once
In my changes, similar to the previous implementation, the priority order for fields being initialized is:
1. If the `init` argument is a tuple, the nth element corresponding to the nth field of the struct.
2. Otherwise, if the `init` argument is not a tuple, try to find a matching field name on `init` and use that field.
3. Is the field has a default value, initalize with that value.
4. Fall back to what the field would have been initialized to via a recursive call to `std.mem.zeroInit`.
But instead of initializing a default instance of the struct and then running multiple passes over it, the init method is chosen per-field and each field is initialized exactly once.
* Initialize `big_align` with 1 as 0 is not a valid alignment.
* Add an assert to `alignForwardGeneric` to catch this issue earlier.
* Refactor valid alignment checks to call a more descriptive function.
* Improve and remove duplicate doNotOptimizeAway() implementations
We currently have two doNotOptimizeAway() implementations, one in
std.math and the other one in std.mem.
Maybe we should deprecate one. In the meantime, the std.math one
now just calls the std.mem one.
In a comptime environment, just ignore the value. Previously,
std.mem.doNotOptimizeAway() did not work at comptime.
If the value fits in a CPU register, just tell the compiler we
need that value to be computed, without clobbering anything else.
Only clobber all possibly escaped memory on pointers or large arrays.
Add tests by the way since we didn't had any (we had, but only
indirect ones).
There are still a few occurrences of "stage1" in the standard library
and self-hosted compiler source, however, these instances need a bit
more careful inspection to ensure no breakage.
Instead of making the memory alignment functions more complicated, I
added more API documentation for their existing semantics.
closes#12118closes#12135
* mem: refactor tests of split()
- add a few cases for .rest()
- use expectEqualSlices()
* mem: add splitBackwards
Over the last couple of weeks weeks I needed to iterate over a
collection backwards at least twice. Do we want to have this in stdlib?
If yes, click "Merge" and start using today! Free shipping and returns
(before 1.0).
Why is this useful?
-------------------
I need this for building an error wrapper: errors are added in the
wrapper from "lowest" level to "highest" level, and then printed in
reverse order. Imagine `UpdateUsers` call, which needs to return
`error.InvalidInput` and a wrappable error context. In Go we would add a
context to the error when returning it:
// if update_user fails, add context on which user we are operating
if err := update_user(user); err != nil {
return fmt.Errorf("user id=%d: %w", user.id, err)
}
Since Zig cannot pass anything else than u16 with an error (#2647), I
will pass a `err_ctx: *Err`, to the callers, where they can, besides
returning an error, augment it with auxiliary data. `Err` is a
preallocated array that can add zero-byte-separated strings. For a
concrete example, imagine such a call graph:
update_user(User, *Err) error{InvalidInput}!<...>
validate_user([]const u8, *Err) error{InvalidInput}!<...>
Where `validate_user` would like, besides only the error, signal the
invalid field. And `update_user`, besides the error, would signal the
offending user id.
We also don't want the low-level functions to know in which context they
are operating to construct a meaningful error message: if validation
fails, they append their "context" to the buffer. To translate/augment
the Go example above:
pub fn validate_user(err_ctx: *Err, user: User) error{InvalidInput}!void {
const name = user.name;
if (!ascii.isAlpha(name)) {
err_ctx.print("name '{s}' must be ascii-letters only", .{name});
return error.InvalidInput;
}
<...>
}
// update_user validates each user and does something with it.
pub fn update_user(err_ctx: *Err, user: User) error{InvalidInput}!void {
// validate the user before updating it
validate_user(user) catch {
err_ctx.print("user id={d}", .{user.id});
return error.InvalidInput;
};
<...>
}
Then the top-level function (in my case, CLI) will read the buffer
backwards (splitting on `"\x00"`) and print:
user id=123: name 'Žvangalas' must be ascii-letters only
To read that buffer backwards, dear readers of this commit message, I
need `mem.splitBackwards`.
Rename all references of sparcv9 to sparc64, to make Zig align more with
other projects. Also, added new function to convert glibc arch name to Zig
arch name, since it refers to the architecture as sparcv9.
This is based on the suggestion by @kubkon in PR 11847.
(https://github.com/ziglang/zig/pull/11487#pullrequestreview-963761757)
