io: soften ‘at most one write attempt’ requirement in io::Write::write #107200

mina86 · 2023-01-22T16:18:15Z

At the moment, documentation of std::io::Write::write indicates that
call to it ‘represents at most one attempt to write to any wrapped
object’. It seems that such wording was put there to contrast it with
pre-1.0 interface which attempted to write all the data (it has since
been changed in RFC 517).

However, the requirement puts unnecessary constraints and may
complicate adaptors which perform non-trivial transformations on the
data. For example, they may maintain an internal buffer which needs
to be written out before the write method accepts more data. It might
be natural to code the method such that it flushes the buffer and then
grabs another chunk of user data. With the current wording in the
documentation, the adaptor would be forced to return Ok(0).

This commit softens the wording such that implementations can choose
code structure which makes most sense for their particular use case.

While at it, elaborate on the meaning of Ok(0) return pointing out
that the write_all methods interprets it as an error.

rustbot · 2023-01-22T16:18:22Z

r? @Mark-Simulacrum

(rustbot has picked a reviewer for you, use r? to override)

rustbot · 2023-01-22T16:18:24Z

Hey! It looks like you've submitted a new PR for the library teams!

If this PR contains changes to any rust-lang/rust public library APIs then please comment with @rustbot label T-libs-api -T-libs to tag it appropriately. If this PR contains changes to any unstable APIs please edit the PR description to add a link to the relevant API Change Proposal or create one if you haven't already. If you're unsure where your change falls no worries, just leave it as is and the reviewer will take a look and make a decision to forward on if necessary.

Examples of T-libs-api changes:

Stabilizing library features
Introducing insta-stable changes such as new implementations of existing stable traits on existing stable types
Introducing new or changing existing unstable library APIs (excluding permanently unstable features / features without a tracking issue)
Changing public documentation in ways that create new stability guarantees
Changing observable runtime behavior of library APIs

mina86 · 2023-01-22T16:19:03Z

@rustbot label T-libs-api -T-libs
cc @alexcrichton

the8472 · 2023-01-22T17:09:24Z

it has since been changed in RFC 517

Which part? searching through the RFC I have found

The biggest change here is to the semantics of write. Instead of repeatedly writing to the underlying IO object until all of buf is written, it attempts a single write and on success returns the number of bytes written.

My understanding is that the at most once guarantee allows transitive atomicity guarantees. If you have a bunch of wrappers (e.g. progress reporting or line-buffering) around File and the underlying OS guarantees atomic append (linux does up to the page size) then you won't get tearing because some intermediate writer misbehaves. This can be important for logging.

Another issue is the return value. Either you get an Ok() with the bytes written or an Err() which indicates the write didn't happen and (for some errors) you can retry it. If some adapter splits the write and writes some bytes and then encounters an error it would have to bubble that up, but then it can't report the bytes. So without this guarantee the caller doesn't know how many bytes it has to retry.

the8472 · 2023-01-22T17:19:27Z

Furthermore, it removes statement that returning Ok(0) means that
the object is likely no longer able to accept data. With adaptor
types this is often untrue and even with system-level file descriptors
I couldn’t find documentation suggesting that write may return zero in
such cases.

The default impl for write_all turns this into an error, so this would be a breaking change.

rust/library/std/src/io/mod.rs

Lines 1539 to 1548 in cef633d

 #[stable(feature = "rust1", since = "1.0.0")] 

 fn write_all(&mut self, mut buf: &[u8]) -> Result<()> { 

 while !buf.is_empty() { 

 match self.write(buf) { 

 Ok(0) => { 

 return Err(error::const_io_error!( 

 ErrorKind::WriteZero, 

 "failed to write whole buffer", 

 )); 

 }

Prior discussion: #56889 (comment)

mina86 · 2023-01-22T20:36:35Z

The biggest change here is to the semantics of write. Instead of
repeatedly writing to the underlying IO object until all of buf is
written, it attempts a single write and on success returns the
number of bytes written.

Yep, that’s the part I was referring to.

My understanding is that the at most once guarantee allows
transitive atomicity guarantees. If you have a bunch of wrappers
around File and the underlying OS guarantees atomic append then
you won't get tearing because some intermediate writer
misbehaves.

But that’s the problem of those intermediate writers. If Linux
guarantees atomic append up to a page, I can easily write an adaptor
conforming to the Write interface which buffers up to two pages.
Requirement of at most one write does not guarantee that atomicity
properties of wrapped objects are preserved.

If some adapter splits the write and writes some bytes and then
encounters an error it would have to bubble that up, but then it
can't report the bytes.

Yes, that adaptor would need to store the error, return
Ok(bytes_written) and return Err(stored_error) on next write. For
some writers that may make sense. The ‘at most once’ guarantee is not
necessary to adhere to other requirements of the Write trait. [edit:
To be clear, I’m not saying writers should loop over input, I’m just
pointing out that ‘at most once’ is not required to uphold other
requirements of the trait.]

Prior discussion: #56889 (comment)

Interesting. I actually stumbled upon this working on base64 as well.
Didn’t notice that Marshall has already brought this up.

I wasn’t aware about the write_all issue so perhaps changing wording
around Ok(0) isn’t appropriate, but then I argue even more strongly
that ‘at most once’ requirement shouldn’t be there. Essentially, I’m
agreeing with @alexcrichton here:

The "at most one write" language I think should be toned down to
a general guideline, as it's basically just saying "please don't
make write equivalent to write_all".

I’ve updated the commit to not change meaning of Ok(0).

At the moment, documentation of std::io::Write::write indicates that call to it ‘represents at most one attempt to write to any wrapped object’. It seems that such wording was put there to contrast it with pre-1.0 interface which attempted to write all the data (it has since been changed in [RFC 517]). However, the requirement puts unnecessary constraints and may complicate adaptors which perform non-trivial transformations on the data. For example, they may maintain an internal buffer which needs to be written out before the write method accepts more data. It might be natural to code the method such that it flushes the buffer and then grabs another chunk of user data. With the current wording in the documentation, the adaptor would be forced to return Ok(0). This commit softens the wording such that implementations can choose code structure which makes most sense for their particular use case. While at it, elaborate on the meaning of `Ok(0)` return pointing out that the write_all methods interprets it as an error. [RFC 517]: https://rust-lang.github.io/rfcs/0517-io-os-reform.html

the8472 · 2023-01-29T18:24:39Z

I think it's a libs-api issue since it changes Trait guarantees.

r? libs-api

the8472 · 2023-01-31T18:30:05Z

I have looked through some previous PR discussions around BufWriter/LineWriter/stdout buffering by @Lucretiel (some of which didn't land). The main aim there was to minimize tearing of written slices, but not a hard guarantee that 1 write call = 1 syscall.

#78551 (comment) argues

It does only try to write the newly submitted data at most once. flush_buf only targets old data.

I think this is somewhat of a middle ground. The new data results in at most one write to the underlying. The old data already "paid" by a previous write call.

Some general edge-cases:

polled IO where the caller needs ErrorKind::WouldBlock to arm epoll&co., storing the error in the writer and only returning it on the next write attempt could be surprising
writes to procfs/sysfs which must be performed in a single syscall. A user may want to println!() into a procfile and use bufwriter to coalesce the writes into a single one
atomic appends
sockets / nagle's algorithm (write-read is fine, write-write-read is bad)

writers that are buffering may flush out previously-acknowledged-and-buffered data at the beginning of a write call. if that errors the error must be propagated immediately
they should (best-effort) avoid tearing a written slice.
- Which means if they're flushing the previously buffered data they should avoid topping up the buffer with a few bytes from the incoming write before flushing it
  - unless the data can be written in one go
    - small write and just-so fits in the buffer
    - is_write_vectored() == true -> write_vectored. avoids nagle
  - sad case: no vectored writes -> buffer flush new write -> write-write-read 😢

-> If we ignore previously written data that's being flushed now then 1 outer write = at most one inner write

mina86 · 2023-01-31T23:43:34Z

polled IO where the caller needs ErrorKind::WouldBlock to arm epoll&co., storing the error in the writer and only returning it on the next write attempt could be surprising

writes to procfs/sysfs which must be performed in a single syscall. A user may want to println!() into a procfile and use bufwriter to coalesce the writes into a single one

atomic appends

sockets / nagle's algorithm (write-read is fine, write-write-read is bad)

None of those work with current Write trait. A Write implementation can store error, write only half of the input or buffer until it has 69KB to write. All of those are valid and will break one of the listed edge cases. This PR doesn’t change anything in that regard. Softening the ‘at most one write’ requirement doesn’t make things worse for the listed cases while at the same time simplifies implementation of whole class of writers (and by simplifies I mean turn them from arduous to straightforward).

[writers] should (best-effort) avoid tearing a written slice.

I don’t think I’d even agree with that phrasing. This is certainly use-case dependent and giving strong preference for particular use-case seems questionable.

joshtriplett · 2023-02-01T04:45:03Z

We talked about this in the @rust-lang/libs-api meeting today.

BufWriter already behaves like this, and we don't probably want to change that (because doing so would add extra copying into the buffer), so we should document the expectation. Given that:

@rfcbot merge

rfcbot · 2023-02-01T04:45:04Z

Team member @joshtriplett has proposed to merge this. The next step is review by the rest of the tagged team members:

No concerns currently listed.

Once a majority of reviewers approve (and at most 2 approvals are outstanding), this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

The wording around Write::write method is changing with requirement that it maps to ‘at most one write’ being removed¹. With that, change EncoderWriter::write so that it flushes entire output buffer at the beginning and then proceeds to process new input. This eliminates returning Ok(0) which is effectively an error. Also, change accounting for the occupied portion of the output buffer. Rather than just having occupied length, track occupied range which means moving data to front is no longer necessary. ¹ rust-lang/rust#107200

io: soften ‘at most one write attempt’ requirement in io::Write::write At the moment, documentation of std::io::Write::write indicates that call to it ‘represents at most one attempt to write to any wrapped object’. It seems that such wording was put there to contrast it with pre-1.0 interface which attempted to write all the data (it has since been changed in [RFC 517]). However, the requirement puts unnecessary constraints and may complicate adaptors which perform non-trivial transformations on the data. For example, they may maintain an internal buffer which needs to be written out before the write method accepts more data. It might be natural to code the method such that it flushes the buffer and then grabs another chunk of user data. With the current wording in the documentation, the adaptor would be forced to return Ok(0). This commit softens the wording such that implementations can choose code structure which makes most sense for their particular use case. While at it, elaborate on the meaning of `Ok(0)` return pointing out that the write_all methods interprets it as an error. [RFC 517]: https://rust-lang.github.io/rfcs/0517-io-os-reform.html

…iaskrgr Rollup of 5 pull requests Successful merges: - rust-lang#107200 (io: soften ‘at most one write attempt’ requirement in io::Write::write) - rust-lang#112667 (Move WF/ConstEvaluatable goal to clause) - rust-lang#112685 (std: only depend on dlmalloc for wasm*-unknown) - rust-lang#112722 (bootstrap: check for dry run when copying env vars for msvc) - rust-lang#112734 (Make `Bound::predicates` use `Clause`) r? `@ghost` `@rustbot` modify labels: rollup

a1phyr · 2023-08-08T18:46:49Z

I think that the change about returning should be reverted, at least for now.

Returning Ok(0) on write used to mean EOF, which is perfectly fine. I don't understand why would the interface change about this.
If writing is not possible now but will be in the future and blocking is not appropriate, returning Err(WouldBlock) is the thing to do.

It is the right thing for write_all to error with Err(WriteZero) on Ok(0), because EOF was reached before the function could do what it was instructed. Now there are two possible errors to check if write_all failed because the writer is full: WriteZero and StorageFull.

Moreover, this proposes StorageFull, which is meant for the underlying storage (eg a disk) more than a writer itself. Also, StorageFull is unstable, which mean most users cannot even respect the new contract.

Additionally, this changes the interface without changing any implementation in std, which continues to return Ok(0) on EOF. Doing so would be a change that actually breaks code so probably won't be accepted. So we end up with std not implementing properly its own traits.

the8472 · 2023-08-08T19:56:04Z

Please open a new issue on that. Most people won't notice discussion on a merged PR.

joshtriplett · 2023-08-16T16:02:11Z

@a1phyr Thanks for catching this! #114897

`Ok(0)` is indeed something the caller may interpret as an error, but that's the *correct* thing to return if the writer can't accept any more bytes.

…r=m-ou-se Partially revert rust-lang#107200 `Ok(0)` is indeed something the caller may interpret as an error, but that's the *correct* thing to return if the writer can't accept any more bytes.

Rollup of 5 pull requests Successful merges: - rust-lang#113715 (Unstable Book: update `lang_items` page and split it) - rust-lang#114897 (Partially revert rust-lang#107200) - rust-lang#114913 (Fix suggestion for attempting to define a string with single quotes) - rust-lang#114931 (Revert PR rust-lang#114052 to fix invalid suggestion) - rust-lang#114944 (update `thiserror` to version >= 1.0.46) r? `@ghost` `@rustbot` modify labels: rollup

`Ok(0)` is indeed something the caller may interpret as an error, but that's the *correct* thing to return if the writer can't accept any more bytes. (cherry picked from commit 5210f48)

[beta] backports * Upgrade std to gimli 0.28.0 rust-lang#114825 * Partially revert rust-lang#107200 rust-lang#114897 * Permit pre-evaluated constants in simd_shuffle rust-lang#113529 r? cuviper

Version 1.72.0 (2023-08-24) ========================== Language -------- - [Replace const eval limit by a lint and add an exponential backoff warning](rust-lang/rust#103877) - [expand: Change how `#![cfg(FALSE)]` behaves on crate root](rust-lang/rust#110141) - [Stabilize inline asm for LoongArch64](rust-lang/rust#111235) - [Uplift `clippy::undropped_manually_drops` lint](rust-lang/rust#111530) - [Uplift `clippy::invalid_utf8_in_unchecked` lint](rust-lang/rust#111543) - [Uplift `clippy::cast_ref_to_mut` lint](rust-lang/rust#111567) - [Uplift `clippy::cmp_nan` lint](rust-lang/rust#111818) - [resolve: Remove artificial import ambiguity errors](rust-lang/rust#112086) - [Don't require associated types with Self: Sized bounds in `dyn Trait` objects](rust-lang/rust#112319) Compiler -------- - [Remember names of `cfg`-ed out items to mention them in diagnostics](rust-lang/rust#109005) - [Support for native WASM exceptions](rust-lang/rust#111322) - [Add support for NetBSD/aarch64-be (big-endian arm64).](rust-lang/rust#111326) - [Write to stdout if `-` is given as output file](rust-lang/rust#111626) - [Force all native libraries to be statically linked when linking a static binary](rust-lang/rust#111698) - [Add Tier 3 support for `loongarch64-unknown-none*`](rust-lang/rust#112310) - [Prevent `.eh_frame` from being emitted for `-C panic=abort`](rust-lang/rust#112403) - [Support 128-bit enum variant in debuginfo codegen](rust-lang/rust#112474) - [compiler: update solaris/illumos to enable tsan support.](rust-lang/rust#112039) Refer to Rust's [platform support page][platform-support-doc] for more information on Rust's tiered platform support. Libraries --------- - [Document memory orderings of `thread::{park, unpark}`](rust-lang/rust#99587) - [io: soften ‘at most one write attempt’ requirement in io::Write::write](rust-lang/rust#107200) - [Specify behavior of HashSet::insert](rust-lang/rust#107619) - [Relax implicit `T: Sized` bounds on `BufReader<T>`, `BufWriter<T>` and `LineWriter<T>`](rust-lang/rust#111074) - [Update runtime guarantee for `select_nth_unstable`](rust-lang/rust#111974) - [Return `Ok` on kill if process has already exited](rust-lang/rust#112863) - [Implement PartialOrd for `Vec`s over different allocators](rust-lang/rust#112632) - [Use 128 bits for TypeId hash](rust-lang/rust#109953) - [Don't drain-on-drop in DrainFilter impls of various collections.](rust-lang/rust#104455) - [Make `{Arc,Rc,Weak}::ptr_eq` ignore pointer metadata](rust-lang/rust#106450) Rustdoc ------- - [Allow whitespace as path separator like double colon](rust-lang/rust#108537) - [Add search result item types after their name](rust-lang/rust#110688) - [Search for slices and arrays by type with `[]`](rust-lang/rust#111958) - [Clean up type unification and "unboxing"](rust-lang/rust#112233) Stabilized APIs --------------- - [`impl<T: Send> Sync for mpsc::Sender<T>`](https://doc.rust-lang.org/nightly/std/sync/mpsc/struct.Sender.html#impl-Sync-for-Sender) - [`impl TryFrom<&OsStr> for &str`](https://doc.rust-lang.org/nightly/std/primitive.str.html#impl-TryFrom<&'a OsStr>-for-&'a str) - [`String::leak`](https://doc.rust-lang.org/nightly/alloc/string/struct.String.html#method.leak) These APIs are now stable in const contexts: - [`CStr::from_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_str`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) Cargo ----- - Enable `-Zdoctest-in-workspace` by default. When running each documentation test, the working directory is set to the root directory of the package the test belongs to. [docs](https://doc.rust-lang.org/nightly/cargo/commands/cargo-test.html#working-directory-of-tests) [openwrt#12221](rust-lang/cargo#12221) [openwrt#12288](rust-lang/cargo#12288) - Add support of the "default" keyword to reset previously set `build.jobs` parallelism back to the default. [openwrt#12222](rust-lang/cargo#12222) Compatibility Notes ------------------- - [Alter `Display` for `Ipv6Addr` for IPv4-compatible addresses](rust-lang/rust#112606) - Cargo changed feature name validation check to a hard error. The warning was added in Rust 1.49. These extended characters aren't allowed on crates.io, so this should only impact users of other registries, or people who don't publish to a registry. [openwrt#12291](rust-lang/cargo#12291) Refreshed patches. Signed-off-by: Tianling Shen <[email protected]>

Version 1.72.0 (2023-08-24) ========================== Language -------- - [Replace const eval limit by a lint and add an exponential backoff warning](rust-lang/rust#103877) - [expand: Change how `#![cfg(FALSE)]` behaves on crate root](rust-lang/rust#110141) - [Stabilize inline asm for LoongArch64](rust-lang/rust#111235) - [Uplift `clippy::undropped_manually_drops` lint](rust-lang/rust#111530) - [Uplift `clippy::invalid_utf8_in_unchecked` lint](rust-lang/rust#111543) - [Uplift `clippy::cast_ref_to_mut` lint](rust-lang/rust#111567) - [Uplift `clippy::cmp_nan` lint](rust-lang/rust#111818) - [resolve: Remove artificial import ambiguity errors](rust-lang/rust#112086) - [Don't require associated types with Self: Sized bounds in `dyn Trait` objects](rust-lang/rust#112319) Compiler -------- - [Remember names of `cfg`-ed out items to mention them in diagnostics](rust-lang/rust#109005) - [Support for native WASM exceptions](rust-lang/rust#111322) - [Add support for NetBSD/aarch64-be (big-endian arm64).](rust-lang/rust#111326) - [Write to stdout if `-` is given as output file](rust-lang/rust#111626) - [Force all native libraries to be statically linked when linking a static binary](rust-lang/rust#111698) - [Add Tier 3 support for `loongarch64-unknown-none*`](rust-lang/rust#112310) - [Prevent `.eh_frame` from being emitted for `-C panic=abort`](rust-lang/rust#112403) - [Support 128-bit enum variant in debuginfo codegen](rust-lang/rust#112474) - [compiler: update solaris/illumos to enable tsan support.](rust-lang/rust#112039) Refer to Rust's [platform support page][platform-support-doc] for more information on Rust's tiered platform support. Libraries --------- - [Document memory orderings of `thread::{park, unpark}`](rust-lang/rust#99587) - [io: soften ‘at most one write attempt’ requirement in io::Write::write](rust-lang/rust#107200) - [Specify behavior of HashSet::insert](rust-lang/rust#107619) - [Relax implicit `T: Sized` bounds on `BufReader<T>`, `BufWriter<T>` and `LineWriter<T>`](rust-lang/rust#111074) - [Update runtime guarantee for `select_nth_unstable`](rust-lang/rust#111974) - [Return `Ok` on kill if process has already exited](rust-lang/rust#112863) - [Implement PartialOrd for `Vec`s over different allocators](rust-lang/rust#112632) - [Use 128 bits for TypeId hash](rust-lang/rust#109953) - [Don't drain-on-drop in DrainFilter impls of various collections.](rust-lang/rust#104455) - [Make `{Arc,Rc,Weak}::ptr_eq` ignore pointer metadata](rust-lang/rust#106450) Rustdoc ------- - [Allow whitespace as path separator like double colon](rust-lang/rust#108537) - [Add search result item types after their name](rust-lang/rust#110688) - [Search for slices and arrays by type with `[]`](rust-lang/rust#111958) - [Clean up type unification and "unboxing"](rust-lang/rust#112233) Stabilized APIs --------------- - [`impl<T: Send> Sync for mpsc::Sender<T>`](https://doc.rust-lang.org/nightly/std/sync/mpsc/struct.Sender.html#impl-Sync-for-Sender) - [`impl TryFrom<&OsStr> for &str`](https://doc.rust-lang.org/nightly/std/primitive.str.html#impl-TryFrom<&'a OsStr>-for-&'a str) - [`String::leak`](https://doc.rust-lang.org/nightly/alloc/string/struct.String.html#method.leak) These APIs are now stable in const contexts: - [`CStr::from_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_str`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) Cargo ----- - Enable `-Zdoctest-in-workspace` by default. When running each documentation test, the working directory is set to the root directory of the package the test belongs to. [docs](https://doc.rust-lang.org/nightly/cargo/commands/cargo-test.html#working-directory-of-tests) [openwrt#12221](rust-lang/cargo#12221) [openwrt#12288](rust-lang/cargo#12288) - Add support of the "default" keyword to reset previously set `build.jobs` parallelism back to the default. [openwrt#12222](rust-lang/cargo#12222) Compatibility Notes ------------------- - [Alter `Display` for `Ipv6Addr` for IPv4-compatible addresses](rust-lang/rust#112606) - Cargo changed feature name validation check to a hard error. The warning was added in Rust 1.49. These extended characters aren't allowed on crates.io, so this should only impact users of other registries, or people who don't publish to a registry. [openwrt#12291](rust-lang/cargo#12291) Refreshed patches. Signed-off-by: Tianling Shen <[email protected]> (cherry picked from commit 846ee0b) Signed-off-by: Jeffery To <[email protected]>

Version 1.72.0 (2023-08-24) ========================== Language -------- - [Replace const eval limit by a lint and add an exponential backoff warning](rust-lang/rust#103877) - [expand: Change how `#![cfg(FALSE)]` behaves on crate root](rust-lang/rust#110141) - [Stabilize inline asm for LoongArch64](rust-lang/rust#111235) - [Uplift `clippy::undropped_manually_drops` lint](rust-lang/rust#111530) - [Uplift `clippy::invalid_utf8_in_unchecked` lint](rust-lang/rust#111543) - [Uplift `clippy::cast_ref_to_mut` lint](rust-lang/rust#111567) - [Uplift `clippy::cmp_nan` lint](rust-lang/rust#111818) - [resolve: Remove artificial import ambiguity errors](rust-lang/rust#112086) - [Don't require associated types with Self: Sized bounds in `dyn Trait` objects](rust-lang/rust#112319) Compiler -------- - [Remember names of `cfg`-ed out items to mention them in diagnostics](rust-lang/rust#109005) - [Support for native WASM exceptions](rust-lang/rust#111322) - [Add support for NetBSD/aarch64-be (big-endian arm64).](rust-lang/rust#111326) - [Write to stdout if `-` is given as output file](rust-lang/rust#111626) - [Force all native libraries to be statically linked when linking a static binary](rust-lang/rust#111698) - [Add Tier 3 support for `loongarch64-unknown-none*`](rust-lang/rust#112310) - [Prevent `.eh_frame` from being emitted for `-C panic=abort`](rust-lang/rust#112403) - [Support 128-bit enum variant in debuginfo codegen](rust-lang/rust#112474) - [compiler: update solaris/illumos to enable tsan support.](rust-lang/rust#112039) Refer to Rust's [platform support page][platform-support-doc] for more information on Rust's tiered platform support. Libraries --------- - [Document memory orderings of `thread::{park, unpark}`](rust-lang/rust#99587) - [io: soften ‘at most one write attempt’ requirement in io::Write::write](rust-lang/rust#107200) - [Specify behavior of HashSet::insert](rust-lang/rust#107619) - [Relax implicit `T: Sized` bounds on `BufReader<T>`, `BufWriter<T>` and `LineWriter<T>`](rust-lang/rust#111074) - [Update runtime guarantee for `select_nth_unstable`](rust-lang/rust#111974) - [Return `Ok` on kill if process has already exited](rust-lang/rust#112863) - [Implement PartialOrd for `Vec`s over different allocators](rust-lang/rust#112632) - [Use 128 bits for TypeId hash](rust-lang/rust#109953) - [Don't drain-on-drop in DrainFilter impls of various collections.](rust-lang/rust#104455) - [Make `{Arc,Rc,Weak}::ptr_eq` ignore pointer metadata](rust-lang/rust#106450) Rustdoc ------- - [Allow whitespace as path separator like double colon](rust-lang/rust#108537) - [Add search result item types after their name](rust-lang/rust#110688) - [Search for slices and arrays by type with `[]`](rust-lang/rust#111958) - [Clean up type unification and "unboxing"](rust-lang/rust#112233) Stabilized APIs --------------- - [`impl<T: Send> Sync for mpsc::Sender<T>`](https://doc.rust-lang.org/nightly/std/sync/mpsc/struct.Sender.html#impl-Sync-for-Sender) - [`impl TryFrom<&OsStr> for &str`](https://doc.rust-lang.org/nightly/std/primitive.str.html#impl-TryFrom<&'a OsStr>-for-&'a str) - [`String::leak`](https://doc.rust-lang.org/nightly/alloc/string/struct.String.html#method.leak) These APIs are now stable in const contexts: - [`CStr::from_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_str`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) Cargo ----- - Enable `-Zdoctest-in-workspace` by default. When running each documentation test, the working directory is set to the root directory of the package the test belongs to. [docs](https://doc.rust-lang.org/nightly/cargo/commands/cargo-test.html#working-directory-of-tests) [#12221](rust-lang/cargo#12221) [#12288](rust-lang/cargo#12288) - Add support of the "default" keyword to reset previously set `build.jobs` parallelism back to the default. [#12222](rust-lang/cargo#12222) Compatibility Notes ------------------- - [Alter `Display` for `Ipv6Addr` for IPv4-compatible addresses](rust-lang/rust#112606) - Cargo changed feature name validation check to a hard error. The warning was added in Rust 1.49. These extended characters aren't allowed on crates.io, so this should only impact users of other registries, or people who don't publish to a registry. [#12291](rust-lang/cargo#12291) Refreshed patches. Signed-off-by: Tianling Shen <[email protected]> (cherry picked from commit 846ee0b) Signed-off-by: Jeffery To <[email protected]>

Pkgsrc changes: * Adjust patches and cargo checksums to new versions. Upstream changes: Version 1.72.0 (2023-08-24) ========================== Language -------- - [Replace const eval limit by a lint and add an exponential backoff warning] (rust-lang/rust#103877) - [expand: Change how `#![cfg(FALSE)]` behaves on crate root] (rust-lang/rust#110141) - [Stabilize inline asm for LoongArch64] (rust-lang/rust#111235) - [Uplift `clippy::undropped_manually_drops` lint] (rust-lang/rust#111530) - [Uplift `clippy::invalid_utf8_in_unchecked` lint] (rust-lang/rust#111543) - [Uplift `clippy::cast_ref_to_mut` lint] (rust-lang/rust#111567) - [Uplift `clippy::cmp_nan` lint] (rust-lang/rust#111818) - [resolve: Remove artificial import ambiguity errors] (rust-lang/rust#112086) - [Don't require associated types with Self: Sized bounds in `dyn Trait` objects] (rust-lang/rust#112319) Compiler -------- - [Remember names of `cfg`-ed out items to mention them in diagnostics] (rust-lang/rust#109005) - [Support for native WASM exceptions] (rust-lang/rust#111322) - [Add support for NetBSD/aarch64-be (big-endian arm64).] (rust-lang/rust#111326) - [Write to stdout if `-` is given as output file] (rust-lang/rust#111626) - [Force all native libraries to be statically linked when linking a static binary] (rust-lang/rust#111698) - [Add Tier 3 support for `loongarch64-unknown-none*`] (rust-lang/rust#112310) - [Prevent `.eh_frame` from being emitted for `-C panic=abort`] (rust-lang/rust#112403) - [Support 128-bit enum variant in debuginfo codegen] (rust-lang/rust#112474) - [compiler: update solaris/illumos to enable tsan support.] (rust-lang/rust#112039) Refer to Rust's [platform support page][platform-support-doc] for more information on Rust's tiered platform support. Libraries --------- - [Document memory orderings of `thread::{park, unpark}`] (rust-lang/rust#99587) - [io: soften â<80><98>at most one write attemptâ<80><99> requirement in io::Write::write] (rust-lang/rust#107200) - [Specify behavior of HashSet::insert] (rust-lang/rust#107619) - [Relax implicit `T: Sized` bounds on `BufReader<T>`, `BufWriter<T>` and `LineWriter<T>`] (rust-lang/rust#111074) - [Update runtime guarantee for `select_nth_unstable`] (rust-lang/rust#111974) - [Return `Ok` on kill if process has already exited] (rust-lang/rust#112863) - [Implement PartialOrd for `Vec`s over different allocators] (rust-lang/rust#112632) - [Use 128 bits for TypeId hash] (rust-lang/rust#109953) - [Don't drain-on-drop in DrainFilter impls of various collections.] (rust-lang/rust#104455) - [Make `{Arc,Rc,Weak}::ptr_eq` ignore pointer metadata] (rust-lang/rust#106450) Rustdoc ------- - [Allow whitespace as path separator like double colon] (rust-lang/rust#108537) - [Add search result item types after their name] (rust-lang/rust#110688) - [Search for slices and arrays by type with `[]`] (rust-lang/rust#111958) - [Clean up type unification and "unboxing"] (rust-lang/rust#112233) Stabilized APIs --------------- - [`impl<T: Send> Sync for mpsc::Sender<T>`] (https://doc.rust-lang.org/nightly/std/sync/mpsc/struct.Sender.html#impl-Sync-for-Sender) - [`impl TryFrom<&OsStr> for &str`] (https://doc.rust-lang.org/nightly/std/primitive.str.html#impl-TryFrom<&'a OsStr>-for-&'a str) - [`String::leak`] (https://doc.rust-lang.org/nightly/alloc/string/struct.String.html#method.leak) These APIs are now stable in const contexts: - [`CStr::from_bytes_with_nul`] (https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes`] (https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes_with_nul`] (https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_str`] (https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) Cargo ----- - Enable `-Zdoctest-in-workspace` by default. When running each documentation test, the working directory is set to the root directory of the package the test belongs to. [docs](https://doc.rust-lang.org/nightly/cargo/commands/cargo-test.html#working-directory-of-tests) [#12221](rust-lang/cargo#12221) [#12288](rust-lang/cargo#12288) - Add support of the "default" keyword to reset previously set `build.jobs` parallelism back to the default. [#12222](rust-lang/cargo#12222) Compatibility Notes ------------------- - [Alter `Display` for `Ipv6Addr` for IPv4-compatible addresses] (rust-lang/rust#112606) - Cargo changed feature name validation check to a hard error. The warning was added in Rust 1.49. These extended characters aren't allowed on crates.io, so this should only impact users of other registries, or people who don't publish to a registry. [#12291](rust-lang/cargo#12291)

Version 1.72.0 (2023-08-24) ========================== Language -------- - [Replace const eval limit by a lint and add an exponential backoff warning](rust-lang/rust#103877) - [expand: Change how `#![cfg(FALSE)]` behaves on crate root](rust-lang/rust#110141) - [Stabilize inline asm for LoongArch64](rust-lang/rust#111235) - [Uplift `clippy::undropped_manually_drops` lint](rust-lang/rust#111530) - [Uplift `clippy::invalid_utf8_in_unchecked` lint](rust-lang/rust#111543) - [Uplift `clippy::cast_ref_to_mut` lint](rust-lang/rust#111567) - [Uplift `clippy::cmp_nan` lint](rust-lang/rust#111818) - [resolve: Remove artificial import ambiguity errors](rust-lang/rust#112086) - [Don't require associated types with Self: Sized bounds in `dyn Trait` objects](rust-lang/rust#112319) Compiler -------- - [Remember names of `cfg`-ed out items to mention them in diagnostics](rust-lang/rust#109005) - [Support for native WASM exceptions](rust-lang/rust#111322) - [Add support for NetBSD/aarch64-be (big-endian arm64).](rust-lang/rust#111326) - [Write to stdout if `-` is given as output file](rust-lang/rust#111626) - [Force all native libraries to be statically linked when linking a static binary](rust-lang/rust#111698) - [Add Tier 3 support for `loongarch64-unknown-none*`](rust-lang/rust#112310) - [Prevent `.eh_frame` from being emitted for `-C panic=abort`](rust-lang/rust#112403) - [Support 128-bit enum variant in debuginfo codegen](rust-lang/rust#112474) - [compiler: update solaris/illumos to enable tsan support.](rust-lang/rust#112039) Refer to Rust's [platform support page][platform-support-doc] for more information on Rust's tiered platform support. Libraries --------- - [Document memory orderings of `thread::{park, unpark}`](rust-lang/rust#99587) - [io: soften ‘at most one write attempt’ requirement in io::Write::write](rust-lang/rust#107200) - [Specify behavior of HashSet::insert](rust-lang/rust#107619) - [Relax implicit `T: Sized` bounds on `BufReader<T>`, `BufWriter<T>` and `LineWriter<T>`](rust-lang/rust#111074) - [Update runtime guarantee for `select_nth_unstable`](rust-lang/rust#111974) - [Return `Ok` on kill if process has already exited](rust-lang/rust#112863) - [Implement PartialOrd for `Vec`s over different allocators](rust-lang/rust#112632) - [Use 128 bits for TypeId hash](rust-lang/rust#109953) - [Don't drain-on-drop in DrainFilter impls of various collections.](rust-lang/rust#104455) - [Make `{Arc,Rc,Weak}::ptr_eq` ignore pointer metadata](rust-lang/rust#106450) Rustdoc ------- - [Allow whitespace as path separator like double colon](rust-lang/rust#108537) - [Add search result item types after their name](rust-lang/rust#110688) - [Search for slices and arrays by type with `[]`](rust-lang/rust#111958) - [Clean up type unification and "unboxing"](rust-lang/rust#112233) Stabilized APIs --------------- - [`impl<T: Send> Sync for mpsc::Sender<T>`](https://doc.rust-lang.org/nightly/std/sync/mpsc/struct.Sender.html#impl-Sync-for-Sender) - [`impl TryFrom<&OsStr> for &str`](https://doc.rust-lang.org/nightly/std/primitive.str.html#impl-TryFrom<&'a OsStr>-for-&'a str) - [`String::leak`](https://doc.rust-lang.org/nightly/alloc/string/struct.String.html#method.leak) These APIs are now stable in const contexts: - [`CStr::from_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_bytes_with_nul`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) - [`CStr::to_str`](https://doc.rust-lang.org/nightly/std/ffi/struct.CStr.html#method.from_bytes_with_nul) Cargo ----- - Enable `-Zdoctest-in-workspace` by default. When running each documentation test, the working directory is set to the root directory of the package the test belongs to. [docs](https://doc.rust-lang.org/nightly/cargo/commands/cargo-test.html#working-directory-of-tests) [openwrt#12221](rust-lang/cargo#12221) [openwrt#12288](rust-lang/cargo#12288) - Add support of the "default" keyword to reset previously set `build.jobs` parallelism back to the default. [openwrt#12222](rust-lang/cargo#12222) Compatibility Notes ------------------- - [Alter `Display` for `Ipv6Addr` for IPv4-compatible addresses](rust-lang/rust#112606) - Cargo changed feature name validation check to a hard error. The warning was added in Rust 1.49. These extended characters aren't allowed on crates.io, so this should only impact users of other registries, or people who don't publish to a registry. [openwrt#12291](rust-lang/cargo#12291) Refreshed patches. Signed-off-by: Tianling Shen <[email protected]>

rustbot assigned Mark-Simulacrum Jan 22, 2023

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Jan 22, 2023

rustbot added T-libs-api Relevant to the library API team, which will review and decide on the PR/issue. and removed T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Jan 22, 2023

mina86 force-pushed the c branch from 00d98e5 to 5b42696 Compare January 22, 2023 20:36

mina86 force-pushed the c branch from 5b42696 to 55045db Compare January 22, 2023 20:48

mina86 force-pushed the c branch from 55045db to 316742e Compare January 24, 2023 02:04

Mark-Simulacrum assigned the8472 and unassigned Mark-Simulacrum Jan 29, 2023

rustbot assigned joshtriplett and unassigned the8472 Jan 29, 2023

joshtriplett added the I-libs-api-nominated Nominated for discussion during a libs-api team meeting. label Jan 29, 2023

rfcbot added proposed-final-comment-period Proposed to merge/close by relevant subteam, see T-<team> label. Will enter FCP once signed off. disposition-merge This issue / PR is in PFCP or FCP with a disposition to merge it. labels Feb 1, 2023

mina86 mentioned this pull request Feb 1, 2023

EncoderWriter: no longer adhere to ‘at most one write’ marshallpierce/rust-base64#220

Open

compiler-errors mentioned this pull request Jun 18, 2023

Rollup of 7 pull requests #112754

Closed

matthiaskrgr mentioned this pull request Jun 18, 2023

Rollup of 5 pull requests #112755

Merged

bors merged commit 876f00a into rust-lang:master Jun 18, 2023

rustbot added this to the 1.72.0 milestone Jun 18, 2023

mina86 deleted the c branch August 8, 2023 21:56

joshtriplett added a commit to joshtriplett/rust that referenced this pull request Aug 16, 2023

Partially revert rust-lang#107200

5210f48

`Ok(0)` is indeed something the caller may interpret as an error, but that's the *correct* thing to return if the writer can't accept any more bytes.

joshtriplett mentioned this pull request Aug 16, 2023

Partially revert #107200 #114897

Merged

cuviper mentioned this pull request Aug 17, 2023

Rollup of 5 pull requests #114951

Merged

cuviper mentioned this pull request Aug 18, 2023

[beta] backports #114955

Merged

unixzii mentioned this pull request Dec 20, 2023

EncoderWriter::write_all leads to WriteZero error marshallpierce/rust-base64#148

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

io: soften ‘at most one write attempt’ requirement in io::Write::write #107200

io: soften ‘at most one write attempt’ requirement in io::Write::write #107200

mina86 commented Jan 22, 2023 •

edited

Loading

rustbot commented Jan 22, 2023

rustbot commented Jan 22, 2023

mina86 commented Jan 22, 2023

the8472 commented Jan 22, 2023

the8472 commented Jan 22, 2023 •

edited

Loading

mina86 commented Jan 22, 2023 •

edited

Loading

the8472 commented Jan 29, 2023

the8472 commented Jan 31, 2023

mina86 commented Jan 31, 2023

joshtriplett commented Feb 1, 2023

rfcbot commented Feb 1, 2023 •

edited by BurntSushi

Loading

a1phyr commented Aug 8, 2023

the8472 commented Aug 8, 2023

joshtriplett commented Aug 16, 2023

io: soften ‘at most one write attempt’ requirement in io::Write::write #107200

io: soften ‘at most one write attempt’ requirement in io::Write::write #107200

Conversation

mina86 commented Jan 22, 2023 • edited Loading

rustbot commented Jan 22, 2023

rustbot commented Jan 22, 2023

mina86 commented Jan 22, 2023

the8472 commented Jan 22, 2023

the8472 commented Jan 22, 2023 • edited Loading

mina86 commented Jan 22, 2023 • edited Loading

the8472 commented Jan 29, 2023

the8472 commented Jan 31, 2023

mina86 commented Jan 31, 2023

joshtriplett commented Feb 1, 2023

rfcbot commented Feb 1, 2023 • edited by BurntSushi Loading

a1phyr commented Aug 8, 2023

the8472 commented Aug 8, 2023

joshtriplett commented Aug 16, 2023

mina86 commented Jan 22, 2023 •

edited

Loading

the8472 commented Jan 22, 2023 •

edited

Loading

mina86 commented Jan 22, 2023 •

edited

Loading

rfcbot commented Feb 1, 2023 •

edited by BurntSushi

Loading