Rollup merge of #89506 - yaymukund:docblock-headings, r=GuillaumeGomez
librustdoc: Use correct heading levels.
Closes #89309
This fixes the `<h#>` header tags throughout the docs to reflect a semantic hierarchy.
- I ran a script to manually check that we don't have any files with multiple `<h1>` tags.
- Also checked that we never incorrectly nest e.g. a `<h2>` under an `<h3>`.
- I also spot-checked a bunch of pages (`trait.Read`, `enum.Ordering`, `primitive.isize`, `trait.Iterator`).
Rollup merge of #89501 - Aaron1011:escaping-name-regions, r=davidtwco
Note specific regions involved in 'borrowed data escapes' error
Fixes #67007
Currently, a 'borrowed data escapes' error does not mention
the specific lifetime involved (except indirectly through a suggestion
about adding a lifetime bound). We now explain the specific lifetime
relationship that failed to hold, which improves otherwise vague
error messages.
This PR renames `std::thread::available_conccurrency` to `std::thread::available_parallelism`.
## Rationale
The API was initially named `std::thread::hardware_concurrency`, mirroring the [C++ API of the same name](https://en.cppreference.com/w/cpp/thread/thread/hardware_concurrency). We eventually decided to omit any reference to the word "hardware" after [this comment](https://github.com/rust-lang/rust/pull/74480#issuecomment-662045841). And so we ended up with `available_concurrency` instead.
---
For a talk I was preparing this week I was reading through ["Understanding and expressing scalable concurrency" (A. Turon, 2013)](http://aturon.github.io/academic/turon-thesis.pdf), and the following passage stood out to me (emphasis mine):
> __Concurrency is a system-structuring mechanism.__ An interactive system that deals with disparate asynchronous events is naturally structured by division into concurrent threads with disparate responsibilities. Doing so creates a better fit between problem and solution, and can also decrease the average latency of the system by preventing long-running computations from obstructing quicker ones.
> __Parallelism is a resource.__ A given machine provides a certain capacity for parallelism, i.e., a bound on the number of computations it can perform simultaneously. The goal is to maximize throughput by intelligently using this resource. For interactive systems, parallelism can decrease latency as well.
_Chapter 2.1: Concurrency is not Parallelism. Page 30._
---
_"Concurrency is a system-structuring mechanism. Parallelism is a resource."_ — It feels like this accurately captures the way we should be thinking about these APIs. What this API returns is not "the amount of concurrency available to the program" which is a property of the program, and thus even with just a single thread is effectively unbounded. But instead it returns "the amount of _parallelism_ available to the program", which is a resource hard-constrained by the machine's capacity (and can be further restricted by e.g. operating systems).
That's why I'd like to propose we rename this API from `available_concurrency` to `available_parallelism`. This still meets the criteria we previously established of not attempting to define what exactly we mean by "hardware", "threads", and other such words. Instead we only talk about "concurrency" as an abstract resource available to our program.
Rollup merge of #89245 - DeveloperC286:iter_mut_fields_to_private, r=joshtriplett
refactor: make VecDeque's IterMut fields module-private, not just crate-private
Made the fields of VecDeque's IterMut private by creating a IterMut::new(...) function to create a new instance of IterMut and migrating usage to use IterMut::new(...).
Rollup merge of #89050 - DeveloperC286:drain_fields_to_private, r=joshtriplett
refactor: VecDeques Drain fields to private
Made the fields of VecDeque's Drain private by creating a Drain::new(...) function to create a new instance of Drain and migrating usage to use Drain::new(...).
I intend these changes to be helpful to readers who are not yet familiar with the quirks of floating-point numbers. Additionally, I felt it was misleading to describe `Nan` as being the result of division by zero, since most divisions by zero (except for 0/0) produce `Infinite` floats, so I moved that remark to the `Infinite` variant with adjustment.
The first sentence of the `Nan` documentation is copied from `f32`; I followed the example of the `f64` documentation by referring to `f32` for general concepts, rather than duplicating the text.
----
I considered making similar changes to the documentation of the `is_*` methods of floats, but decided that that was a much larger and trickier problem; here, each of the variants' descriptions can be expected to be read in context of being mutually exclusive with the others.
bors [Wed, 6 Oct 2021 06:20:25 +0000 (06:20 +0000)]
Auto merge of #89323 - estebank:derive-binop, r=petrochenkov
Consider unfulfilled obligations in binop errors
When encountering a binop where the types would have been accepted, if
all the predicates had been fulfilled, include information about the
predicates and suggest appropriate `#[derive]`s if possible.
When encountering a binop where the types would have been accepted, if
all the predicates had been fulfilled, include information about the
predicates and suggest appropriate `#[derive]`s if possible.
bors [Tue, 5 Oct 2021 22:28:40 +0000 (22:28 +0000)]
Auto merge of #89572 - Manishearth:rollup-obz5ycp, r=Manishearth
Rollup of 10 pull requests
Successful merges:
- #88706 (Normalize associated type projections when checking return type of main)
- #88828 (Use `libc::sigaction()` instead of `sys::signal()` to prevent a deadlock)
- #88871 (Fix suggestion for nested struct patterns)
- #89317 (Move generic error message to separate branches)
- #89351 (for signed wrapping remainder, do not compare lhs with MIN)
- #89442 (Add check for duplicated doc aliases)
- #89502 (Fix Lower/UpperExp formatting for integers and precision zero)
- #89523 (Make `proc_macro_derive_resolution_fallback` a future-breakage lint)
- #89532 (Document behavior of `MaybeLiveLocals` regarding enums and field-senstivity)
- #89546 (Make an initial guess for metadata size to reduce buffer resizes)
Rollup merge of #89546 - joshtriplett:grow-metadata-faster, r=petrochenkov
Make an initial guess for metadata size to reduce buffer resizes
When reading metadata, the compiler starts with a `Vec::new()`, which will need to grow repeatedly as the metadata gets decompressed into it. Reduce the number of resizes by starting out at the size of the compressed data.
Rollup merge of #89532 - ecstatic-morse:maybe-live-locals-enum, r=oli-obk,tmiasko
Document behavior of `MaybeLiveLocals` regarding enums and field-senstivity
This arose from a [discussion on Zulip](https://rust-lang.zulipchat.com/#narrow/stream/189540-t-compiler.2Fwg-mir-opt/topic/MaybeLiveLocals.20and.20Discriminants) where a new contributor attempted to implement a dead-store elimination pass using this analysis. They ran into a nasty hack around `SetDiscriminant` the effect of which is to lets handle assignments of literals to enum-typed locals (e.g. `x = Some(4)`) correctly. This took me a while to figure out.
Document this oddity, so the next person will have an easier time, and add a test to enshrine the current behavior.
Rollup merge of #89523 - Aaron1011:derive-future-compat, r=wesleywiser
Make `proc_macro_derive_resolution_fallback` a future-breakage lint
When `cargo report future-incompatibilities` is stabilized
(see #71249), this will cause dependencies that trigger
this lint to be included in the report.
Rollup merge of #89502 - FabianWolff:issue-89493, r=joshtriplett
Fix Lower/UpperExp formatting for integers and precision zero
Fixes the integer part of #89493 (I daren't touch the floating-point formatting code). The issue is that the "subtracted" precision essentially behaves like extra trailing zeros, but this is not currently reflected in the code properly.
Rollup merge of #88828 - FabianWolff:issue-88585, r=dtolnay
Use `libc::sigaction()` instead of `sys::signal()` to prevent a deadlock
Fixes #88585. POSIX [specifies](https://man7.org/linux/man-pages/man3/fork.3p.html) that after forking,
> to avoid errors, the child process may only execute async-signal-safe operations until such time as one of the exec functions is called.
Rust's standard library does not currently adhere to this, as evidenced by #88585. The child process calls [`sys::signal()`](https://github.com/rust-lang/rust/blob/7bf0736e130e2203c58654f7353dbf9575e49d5c/library/std/src/sys/unix/android.rs#L76), which on Android calls [`libc::dlsym()`](https://github.com/rust-lang/rust/blob/7bf0736e130e2203c58654f7353dbf9575e49d5c/library/std/src/sys/unix/weak.rs#L101), which is [**not**](https://man7.org/linux/man-pages/man7/signal-safety.7.html) async-signal-safe, and in fact causes a deadlock in the example in #88585.
I think the easiest solution here would be to just call `libc::sigaction()` instead, which [is](https://man7.org/linux/man-pages/man7/signal-safety.7.html) async-signal-safe, provides the functionality we need, and is apparently available on all Android versions because it is also used e.g. [here](https://github.com/rust-lang/rust/blob/7bf0736e130e2203c58654f7353dbf9575e49d5c/library/std/src/sys/unix/stack_overflow.rs#L112-L114).
Rollup merge of #88706 - ThePuzzlemaker:issue-88609, r=jackh726
Normalize associated type projections when checking return type of main
This fixes #88609.
Previously, the return type of `fn main()` would not have any associated type projections within normalized before checking if it implements the standard library trait `std::process::Termination`. This commit appears to fix it.
This feels vaguely symptomatic of a problem in the underlying trait solving engine, but I am not sure how I would solve that. I am unsure why the example in #88609 with `assert_impl_termination` and `fn foo()` work, but simply `fn main()` doesn't. The way that I solved this is also probably not the best way to do this, so please let me know if there is a better way to do this.
I have added a build-pass regression test for this issue.
Aaron Hill [Sun, 3 Oct 2021 20:25:26 +0000 (15:25 -0500)]
Note specific regions involved in 'borrowed data escapes' error
Fixes #67007
Currently, a 'borrowed data escapes' error does not mention
the specific lifetime involved (except indirectly through a suggestion
about adding a lifetime bound). We now explain the specific lifetime
relationship that failed to hold, which improves otherwise vague
error messages.
Made the fields of VecDeque's IterMut private by creating a IterMut::new(...) function to create a new instance of IterMut and migrating usage to use IterMut::new(...).
Trevor Spiteri [Tue, 5 Oct 2021 13:15:24 +0000 (15:15 +0200)]
for signed overflowing remainder, delay comparing lhs with MIN
Since the wrapped remainder is going to be 0 for all cases when the rhs is -1,
there is no need to divide in this case. Comparing the lhs with MIN is only done
for the overflow bool. In particular, this results in better code generation for
wrapping remainder, which discards the overflow bool completely.
bors [Tue, 5 Oct 2021 09:45:11 +0000 (09:45 +0000)]
Auto merge of #89266 - cjgillot:session-ich, r=michaelwoerister
Move ICH to rustc_query_system
Based on https://github.com/rust-lang/rust/pull/89183
The StableHashingContext does not need to be in rustc_middle.
This PR moves it to rustc_query_system. This will avoid a dependency between rustc_ast_lowering and rustc_middle in https://github.com/rust-lang/rust/pull/89124.
bors [Tue, 5 Oct 2021 06:56:57 +0000 (06:56 +0000)]
Auto merge of #89549 - Manishearth:rollup-mhkyc16, r=Manishearth
Rollup of 12 pull requests
Successful merges:
- #87631 (os current_exe using same approach as linux to get always the full ab…)
- #88234 (rustdoc-json: Don't ignore impls for primitive types)
- #88651 (Use the 64b inner:monotonize() implementation not the 128b one for aarch64)
- #88816 (Rustdoc migrate to table so the gui can handle >2k constants)
- #89244 (refactor: VecDeques PairSlices fields to private)
- #89364 (rustdoc-json: Encode json files with UTF-8)
- #89423 (Fix ICE caused by non_exaustive_omitted_patterns struct lint)
- #89426 (bootstrap: add config option for nix patching)
- #89462 (haiku thread affinity build fix)
- #89482 (Follow the diagnostic output style guide)
- #89504 (Don't suggest replacing region with 'static in NLL)
- #89535 (fix busted JavaScript in error index generator)
Rollup merge of #89535 - notriddle:notriddle/error-index-generator-js, r=Mark-Simulacrum
fix busted JavaScript in error index generator
The old JavaScript didn't work. It filled the browser console with "e.previousElementSibling not defined" errors, because it didn't account for the example-wrap div that a newer version of rustdoc added.
Additionally, it had copied versions of utility functions that had been optimized in rustdoc main.js. This version updates those.
Rollup merge of #89504 - Aaron1011:rpit-nll-static, r=nikomatsakis
Don't suggest replacing region with 'static in NLL
Fixes #73159
This is similar to #69350 - if the user didn't initially
write out a 'static lifetime, adding 'static in response to
a lifetime error is usually the wrong thing to do.
Rollup merge of #89426 - davidtwco:bootstrap-nix-toolchain-env-var, r=Mark-Simulacrum
bootstrap: add config option for nix patching
On NixOS systems, bootstrap will patch rustc used in bootstrapping after checking `/etc/os-release` (to confirm the current distribution is NixOS). However, when using Nix on a non-NixOS system, it can be desirable for bootstrap to patch rustc. In this commit, a `patch-binaries-for-nix` option is added to `config.toml`, which allows for user opt-in to bootstrap's Nix patching.
Rollup merge of #88816 - dns2utf8:rustdoc_test_gui_2k_constants, r=GuillaumeGomez
Rustdoc migrate to table so the gui can handle >2k constants
Closes #88545.
This PR adds a test for overlapping entries in the `item-table` https://github.com/rust-lang/rust/issues/88545
It currently includes the commit with the workaround from https://github.com/rust-lang/rust/pull/88776
Rollup merge of #88651 - AGSaidi:monotonize-inner-64b-aarch64, r=dtolnay
Use the 64b inner:monotonize() implementation not the 128b one for aarch64
aarch64 prior to v8.4 (FEAT_LSE2) doesn't have an instruction that guarantees
untorn 128b reads except for completing a 128b load/store exclusive pair
(ldxp/stxp) or compare-and-swap (casp) successfully. The requirement to
complete a 128b read+write atomic is actually more expensive and more unfair
than the previous implementation of monotonize() which used a Mutex on aarch64,
especially at large core counts. For aarch64 switch to the 64b atomic
implementation which is about 13x faster for a benchmark that involves many
calls to Instant::now().
Rollup merge of #88234 - hkmatsumoto:rustdoc-impls-for-primitive, r=jyn514
rustdoc-json: Don't ignore impls for primitive types
Fix the issue discussed at [Zulip](https://rust-lang.zulipchat.com/#narrow/stream/266220-rustdoc/topic/.E2.9C.94.20Json.20output.20lacks.20some.20item.20which.20are.20supposed.20to.20be.20there)
bors [Tue, 5 Oct 2021 04:13:35 +0000 (04:13 +0000)]
Auto merge of #89545 - workingjubilee:rollup-ooxf3p2, r=workingjubilee
Rollup of 15 pull requests
Successful merges:
- #87993 (Stabilize try_reserve)
- #88090 (Perform type inference in range pattern)
- #88780 (Added abs_diff for integer types.)
- #89270 (path.push() should work as expected on windows verbatim paths)
- #89413 (Correctly handle supertraits for min_specialization)
- #89456 (Update to the final LLVM 13.0.0 release)
- #89466 (Fix bug with query modifier parsing)
- #89473 (Fix extra `non_snake_case` warning for shorthand field bindings)
- #89474 (rustdoc: Improve doctest pass's name and module's name)
- #89478 (Fixed numerus of error message)
- #89480 (Add test for issue 89118.)
- #89487 (Try to recover from a `=>` -> `=` or `->` typo in a match arm)
- #89494 (Deny `where` clauses on `auto` traits)
- #89511 (:arrow_up: rust-analyzer)
- #89536 (update Miri)
Jubilee [Tue, 5 Oct 2021 04:12:40 +0000 (21:12 -0700)]
Rollup merge of #89478 - zvavybir:master, r=jyn514
Fixed numerus of error message
When there are redundant trait requirements and these are hidden, a message is generated by the following code snippet:
`format!("{} redundant requirements hidden", count)`
But if there is only a single hidden requirement, it will still print this message in plural instead of singular.
Jubilee [Tue, 5 Oct 2021 04:12:38 +0000 (21:12 -0700)]
Rollup merge of #89473 - FabianWolff:issue-89469, r=joshtriplett
Fix extra `non_snake_case` warning for shorthand field bindings
Fixes #89469. The problem is the innermost `if` condition here:
https://github.com/rust-lang/rust/blob/d14731cb3ced8318d7fc83cbe838f0e7f2fb3b40/compiler/rustc_lint/src/nonstandard_style.rs#L435-L452
This code runs for every `PatKind::Binding`, so if a struct has multiple fields, say A and B, and both are bound in a pattern using shorthands, the call to `self.check_snake_case()` will indeed be skipped in the `check_pat()` call for `A`; but when `check_pat()` is called for `B`, the loop will still iterate over `A`, and `field.ident (= A) != ident (= B)` will be true. I have fixed this by only looking at non-shorthand bindings, and only the binding that `check_pat()` was actually called for.
Jubilee [Tue, 5 Oct 2021 04:12:37 +0000 (21:12 -0700)]
Rollup merge of #89466 - Mark-Simulacrum:query-macros, r=oli-obk
Fix bug with query modifier parsing
The previous macro_rules! parsers failed when an additional modifier was added
with ambiguity errors. The error is pretty unclear as to what exactly the cause
here is, but this change simplifies the argument parsing code such that the
error is avoided.
Extracted from other work, and somewhat duplicates 0358edeb5 from #85830, but
this approach seems a little simpler to me. Not technically currently necessary but seems
like a good cleanup.
Jubilee [Tue, 5 Oct 2021 04:12:35 +0000 (21:12 -0700)]
Rollup merge of #89270 - seanyoung:join_fold, r=m-ou-se
path.push() should work as expected on windows verbatim paths
On Windows, std::fs::canonicalize() returns an so-called UNC path. UNC paths differ with regular paths because:
- This type of path can much longer than a non-UNC path (32k vs 260 characters).
- The prefix for a UNC path is ``Component::Prefix(Prefix::DiskVerbatim(..)))``
- No `/` is allowed
- No `.` is allowed
- No `..` is allowed
Rust has poor handling of such paths. If you join a UNC path with a path with any of the above, then this will not work.
I've implemented a new method `fn join_fold()` which joins paths and also removes any `.` and `..` from it, and replaces `/` with `\` on Windows. Using this function it is possible to use UNC paths without issue. In addition, this function is useful on Linux too; paths can be appended without having to call `canonicalize()` to remove the `.` and `..`.
This PR needs test cases, which can I add. I hope this will a start of a discussion.
Michael Howell [Mon, 4 Oct 2021 20:01:50 +0000 (13:01 -0700)]
fix busted JavaScript in error index generator
The old JavaScript didn't work. It filled the browser console
with "e.previousElementSibling not defined" errors, because
it didn't account for the example-wrap div that a newer version
of rustdoc added.
Additionally, it had copied versions of utility functions that
had been optimized in rustdoc main.js. This version updates those.
Jubilee [Mon, 4 Oct 2021 20:58:16 +0000 (13:58 -0700)]
Rollup merge of #89500 - FabianWolff:issue-87308, r=petrochenkov
Fix ICE with buffered lint referring to AST node deleted by everybody_loops
Fixes #87308. Note the following comment:
https://github.com/rust-lang/rust/blob/08759c691e2e9799a3c6780ffdf910240ebd4a6b/compiler/rustc_lint/src/early.rs#L415-L417
As it turns out, this is not _always_ a bug, because `-Zunpretty=everybody_loops` causes a lot of AST nodes to be deleted, and thus some buffered lints will refer to non-existent node ids. To fix this, my changes simply ignore buffered lints if `-Zunpretty=everybody_loops` is enabled, which, from my understanding, shouldn't be a big issue because it only affects pretty-printing. Of course, a more elegant solution would only ignore buffered lints that actually point at deleted node ids, but I haven't figured out an easy way of achieving this.
For the concrete example in #87308, the buffered lint is created [here](https://github.com/rust-lang/rust/blob/08759c691e2e9799a3c6780ffdf910240ebd4a6b/compiler/rustc_expand/src/mbe/macro_rules.rs#L145-L151) with the `lint_node_id` from [here](https://github.com/rust-lang/rust/blob/08759c691e2e9799a3c6780ffdf910240ebd4a6b/compiler/rustc_expand/src/mbe/macro_rules.rs#L319), i.e. it points at the macro _expansion_, which then gets deleted by `ReplaceBodyWithLoop` [here](https://github.com/rust-lang/rust/blob/08759c691e2e9799a3c6780ffdf910240ebd4a6b/compiler/rustc_interface/src/passes.rs#L377).
Jubilee [Mon, 4 Oct 2021 20:58:14 +0000 (13:58 -0700)]
Rollup merge of #89453 - waywardmonkeys:consistent-supertrait-usage, r=nagisa
Consistently use 'supertrait'.
A subset of places referred to 'super-trait', so this changes them
to all use 'supertrait'. This matches 'supertype' and some other
usages. An exception is 'auto-trait' which is consistently used
in that manner.
Jubilee [Mon, 4 Oct 2021 20:58:07 +0000 (13:58 -0700)]
Rollup merge of #88451 - theo-lw:issue-87771, r=jackh726
Fix an ICE caused by type mismatch errors being ignored
This PR fixes #87771. It turns out that the check on `compiler/rustc_typeck/src/check/demand.rs:148` leads to the ICE. I removed it because the early return in [`check_expr_assign`](https://github.com/theo-lw/rust/blob/dec7fc3ced5bc3c18d0e5d29921d087f93189cb8/compiler/rustc_typeck/src/check/expr.rs#L928) already prevents unnecessary error messages from the call to `check_expr_coercable_to_type`.
Jubilee [Mon, 4 Oct 2021 20:58:07 +0000 (13:58 -0700)]
Rollup merge of #87091 - the8472:more-advance-by-impls, r=joshtriplett
implement advance_(back_)_by on more iterators
Add more efficient, non-default implementations for `feature(iter_advance_by)` (#77404) on more iterators and adapters.
This PR only contains implementations where skipping over items doesn't elide any observable side-effects such as user-provided closures or `clone()` functions. I'll put those in a separate PR.
Jubilee [Mon, 4 Oct 2021 20:58:06 +0000 (13:58 -0700)]
Rollup merge of #83655 - sebpop:arm64-outline-atomics, r=workingjubilee
[aarch64] add target feature outline-atomics
Enable outline-atomics by default as enabled in clang by the following commit
https://reviews.llvm.org/rGc5e7e649d537067dec7111f3de1430d0fc8a4d11
Performance improves by several orders of magnitude when using the LSE instructions
instead of the ARMv8.0 compatible load/store exclusive instructions.
Tested on Graviton2 aarch64-linux with
x.py build && x.py install && x.py test
Aaron Hill [Mon, 4 Oct 2021 16:05:48 +0000 (11:05 -0500)]
Make `proc_macro_derive_resolution_fallback` a future-breakage lint
When `cargo report future-incompatibilities` is stabilized
(see #71249), this will cause dependencies that trigger
this lint to be included in the report.
bors [Mon, 4 Oct 2021 12:49:57 +0000 (12:49 +0000)]
Auto merge of #88834 - the8472:char-count, r=joshtriplett
optimize str::from_utf8() validation when slice contains multibyte chars and str.chars().count() in all cases
The change shows small but consistent improvements across several x86 target feature levels. I also tried to optimize counting with `slice.as_chunks` but that yielded more inconsistent results, bigger improvements for some optimization levels, lesser ones in others.
```
old, -O2, x86-64
test str::str_char_count_emoji ... bench: 1,924 ns/iter (+/- 26)
test str::str_char_count_lorem ... bench: 879 ns/iter (+/- 12)
test str::str_char_count_lorem_short ... bench: 5 ns/iter (+/- 0)
new, -O2, x86-64
test str::str_char_count_emoji ... bench: 1,878 ns/iter (+/- 21)
test str::str_char_count_lorem ... bench: 851 ns/iter (+/- 11)
test str::str_char_count_lorem_short ... bench: 4 ns/iter (+/- 0)
old, -O2, x86-64-v2
test str::str_char_count_emoji ... bench: 1,477 ns/iter (+/- 46)
test str::str_char_count_lorem ... bench: 675 ns/iter (+/- 15)
test str::str_char_count_lorem_short ... bench: 5 ns/iter (+/- 0)
new, -O2, x86-64-v2
test str::str_char_count_emoji ... bench: 1,323 ns/iter (+/- 39)
test str::str_char_count_lorem ... bench: 593 ns/iter (+/- 18)
test str::str_char_count_lorem_short ... bench: 4 ns/iter (+/- 0)
old, -O2, x86-64-v3
test str::str_char_count_emoji ... bench: 748 ns/iter (+/- 7)
test str::str_char_count_lorem ... bench: 348 ns/iter (+/- 2)
test str::str_char_count_lorem_short ... bench: 5 ns/iter (+/- 0)
new, -O2, x86-64-v3
test str::str_char_count_emoji ... bench: 650 ns/iter (+/- 4)
test str::str_char_count_lorem ... bench: 301 ns/iter (+/- 1)
test str::str_char_count_lorem_short ... bench: 5 ns/iter (+/- 0)
```