Skip to content

Fixup stout/stderr on Windows #32184

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 3 commits into from
Closed

Fixup stout/stderr on Windows #32184

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
8427efa
Fixup stout/stderr on Windows
ollie27 Mar 10, 2016
6993939
Simplify Windows stdout/stderr
ollie27 Mar 10, 2016
bd80a53
Further simplify Windows stdout/stderr
ollie27 Mar 12, 2016
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 2 additions & 19 deletions src/libstd/io/stdio.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,6 @@ use prelude::v1::*;
use io::prelude::*;

use cell::{RefCell, BorrowState};
use cmp;
use fmt;
use io::lazy::Lazy;
use io::{self, BufReader, LineWriter};
Expand Down Expand Up @@ -312,22 +311,6 @@ impl<'a> BufRead for StdinLock<'a> {
fn consume(&mut self, n: usize) { self.inner.consume(n) }
}

// As with stdin on windows, stdout often can't handle writes of large
// sizes. For an example, see #14940. For this reason, don't try to
// write the entire output buffer on windows. On unix we can just
// write the whole buffer all at once.
//
// For some other references, it appears that this problem has been
// encountered by others [1] [2]. We choose the number 8KB just because
// libuv does the same.
//
// [1]: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1232
// [2]: http://www.mail-archive.com/[email protected]/msg00661.html
#[cfg(windows)]
const OUT_MAX: usize = 8192;
#[cfg(unix)]
const OUT_MAX: usize = ::usize::MAX;

/// A handle to the global standard output stream of the current process.
///
/// Each handle shares a global buffer of data to be written to the standard
Expand Down Expand Up @@ -440,7 +423,7 @@ impl Write for Stdout {
#[stable(feature = "rust1", since = "1.0.0")]
impl<'a> Write for StdoutLock<'a> {
fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
self.inner.borrow_mut().write(&buf[..cmp::min(buf.len(), OUT_MAX)])
self.inner.borrow_mut().write(buf)
}
fn flush(&mut self) -> io::Result<()> {
self.inner.borrow_mut().flush()
Expand Down Expand Up @@ -546,7 +529,7 @@ impl Write for Stderr {
#[stable(feature = "rust1", since = "1.0.0")]
impl<'a> Write for StderrLock<'a> {
fn write(&mut self, buf: &[u8]) -> io::Result<usize> {
self.inner.borrow_mut().write(&buf[..cmp::min(buf.len(), OUT_MAX)])
self.inner.borrow_mut().write(buf)
}
fn flush(&mut self) -> io::Result<()> {
self.inner.borrow_mut().flush()
Expand Down
26 changes: 23 additions & 3 deletions src/libstd/sys/windows/stdio.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,8 +58,28 @@ fn write(out: &Output, data: &[u8]) -> io::Result<usize> {
Output::Console(ref c) => c.get().raw(),
Output::Pipe(ref p) => return p.get().write(data),
};
let utf16 = match str::from_utf8(data).ok() {
Some(utf8) => utf8.encode_utf16().collect::<Vec<u16>>(),
// As with stdin on windows, stdout often can't handle writes of large
// sizes. For an example, see #14940. For this reason, don't try to
// write the entire output buffer on windows.
//
// For some other references, it appears that this problem has been
// encountered by others [1] [2]. We choose the number 8K just because
// libuv does the same.
//
// [1]: https://tahoe-lafs.org/trac/tahoe-lafs/ticket/1232
// [2]: http://www.mail-archive.com/[email protected]/msg00661.html
const OUT_MAX: usize = 8192;
let (utf16, data_len) = match str::from_utf8(data).ok() {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Perhaps this could do something like:

let len = cmp::min(data.len(), OUT_MAX);
let data = match str::from_utf8(&data[..len]) {
    Ok(s) => s,
    Err(ref e) if e.valid_up_to() == 0 => return Err(..),
    Err(e) => str::from_utf8(&s[..e.valid_up_to()]).unwrap(),
};
let data = ...;

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In the sense that it would:

  • Be the same performance for the "fast path"
  • Handle writing partially utf-8 data naturally
  • Not have to encode a let's-go-backwards loop

Some(mut utf8) => {
if utf8.len() > OUT_MAX {
let mut new_len = OUT_MAX;
while !utf8.is_char_boundary(new_len) {
new_len -= 1;
}
utf8 = &utf8[..new_len];
}
(utf8.encode_utf16().collect::<Vec<u16>>(), utf8.len())
}
None => return Err(invalid_encoding()),
};
let mut written = 0;
Expand All @@ -74,7 +94,7 @@ fn write(out: &Output, data: &[u8]) -> io::Result<usize> {
// FIXME if this only partially writes the utf16 buffer then we need to
// figure out how many bytes of `data` were actually written
assert_eq!(written as usize, utf16.len());
Ok(data.len())
Ok(data_len)
}

impl Stdin {
Expand Down