1. 24 Dec, 2018 4 commits
  2. 30 Nov, 2018 1 commit
    • Alan Somers's avatar
      Clippy cleanup (#889) · 642e56db
      Alan Somers authored
      Prefer pass-by-value over pass-by-reference, but only for private
      functions.  Suppress the warning generally, because it's triggered by
      public functions that can't be changed backwards-compatibly.
      642e56db
  3. 21 Aug, 2018 1 commit
  4. 03 Jul, 2018 1 commit
  5. 08 Mar, 2018 1 commit
  6. 07 Mar, 2018 1 commit
    • Carl Lerche's avatar
      Add `Ready::all` as well as `usize` conversions. (#825) · 6bdfdba0
      Carl Lerche authored
      This patch provides a new constructor for `Ready` that returns a
      value representing readiness for all operations. This is useful for
      registering interest on all operations or for using the value as a mask.
      
      This patch also provides `usize` conversions for `Ready`. This allows
      storing and loading a `Ready` value in an `AtomicUsize`.
      6bdfdba0
  7. 26 Feb, 2018 1 commit
  8. 30 Jan, 2018 1 commit
    • Carl Lerche's avatar
      TcpListener::from_std does not require the addr (#814) · 21463108
      Carl Lerche authored
      The current API for converting a std TcpListener to a mio TcpListener
      requires passing in a SocketAddr. This is because Windows requires the
      address family when initialzing the listener.
      
      However, the TcpListener already knows its socket address at this point,
      so it should not be a required argument.
      21463108
  9. 26 Jan, 2018 1 commit
  10. 05 Jan, 2018 2 commits
  11. 04 Jan, 2018 1 commit
  12. 14 Dec, 2017 1 commit
  13. 28 Sep, 2017 1 commit
    • Alex Crichton's avatar
      Add TcpListener::accept_std (#733) · 878bee8e
      Alex Crichton authored
      This commit adds an `accept_std` method to `TcpListener` which is added with
      the intention of helping downstream libraries improve interoperability without
      exposing `mio` as a public dependency. The `TcpStream` type is already
      convertible from a raw `std::net::TcpStream`, so it should be possible to
      quickly reassociate this socket back with mio.
      878bee8e
  14. 31 Aug, 2017 1 commit
  15. 24 Aug, 2017 1 commit
  16. 23 Aug, 2017 1 commit
  17. 22 Aug, 2017 2 commits
  18. 20 Aug, 2017 2 commits
  19. 17 Aug, 2017 1 commit
  20. 12 Aug, 2017 1 commit
  21. 28 Jul, 2017 1 commit
  22. 27 Jul, 2017 1 commit
    • Iku Iwasa's avatar
      Fix NetBSD build failure due to kevent type difference (#621) · 0cab6caa
      Iku Iwasa authored
      * Add NetBSD to the CI run
      
      * Adapt to kevent type difference on NetBSD
      
      * In struct kevent,
        filter is uint32_t on NetBSD, while short on other platforms.
        And udata is intptr_t on NetBSD, while void * on other platforms.
      
      * In kevent() function,
        nchanges and nevents are size_t on NetBSD, while int on other
        platforms.
      0cab6caa
  23. 17 Jul, 2017 1 commit
    • Taylor Cramer's avatar
      Fuchsia port (#629) · b6dde86e
      Taylor Cramer authored
      * Fuchsia Shell
      
      * Selector
      
      * Working selector and awakener
      
      * A bunch more crap
      
      * Impl UdpSocket and add comments
      
      * UdpSocket fixup
      
      * Initial TCP
      
      * All tests passing
      
      * Start cleanup
      
      * Separate out modules
      
      * Add EventedHandle and fuchsia_channel test
      
      * Remove unused [feature]
      
      * Add only v6 methods and fixup reregistration of EventedFds
      
      * Remove dependency on concurrent_hashmap
      
      * Don't overwrite existing tokens
      
      * panic -> err
      
      * Remove unnecessary pub restrictions and document unsafety
      
      * Fix privacy error and convert panic on invalid token into err
      
      * Use saturating ops for time modification
      
      * Cleanup EventedFd registration
      
      * Replace Events Option with Vec
      
      * Use accessors rather than pub fields
      
      * Clear up unnecessary repeating events on reregistered handles
      
      * Fix orderings
      
      * Cleanup
      
      * Backcompat with rust 1.9
      b6dde86e
  24. 26 Jun, 2017 1 commit
  25. 27 May, 2017 1 commit
    • Carl Lerche's avatar
      Expose TCP socket options (#610) · 000a2e43
      Carl Lerche authored
      This is done by exposing the options provided by net2 directly instead of implementing `TcpStreamExt`. Read & write timeout accessors are omitted given that the sockets are assumed to be in non-blocking mode. These accessors can be added later if they are needed.
      000a2e43
  26. 22 May, 2017 1 commit
  27. 17 May, 2017 1 commit
    • Alan Somers's avatar
      POSIX AIO support (#586) · 3d09a479
      Alan Somers authored
      * POSIX AIO support
      
      On BSD-based operating systems, add low-level support for queuing AIO
      operations in the event loop.  This commit adds 3 new public methods:
      UnixReady::aio, UnixReady::is_aio, and Poll.as_raw_fd.  Full support
      will be added in a separate crate.
      
      * POSIX AIO support
      
      Revert f8a8e4a's changes to ReadinessState.  Add unit tests.
      3d09a479
  28. 29 Apr, 2017 1 commit
  29. 28 Apr, 2017 1 commit
  30. 07 Apr, 2017 1 commit
    • Carl Lerche's avatar
      kqueue: Ignore EPIPE coming out of `kevent` (#584) · fe551aa8
      Carl Lerche authored
      This commit updates mio to ignore `EPIPE` coming out of `kevent`. It
      turns out that older versions of OSX (10.10 and 10.11 have been
      confirmed) will return an `EPIPE` when registering one half of a pipe
      where the other half has been closed. On other platforms this is an ok
      operation which just returns that the pipe is readable/writable (to
      return an error/eof), so this brings the OSX behavior in line by
      ignoring the error and moving to the next event.
      
      Closes #582
      fe551aa8
  31. 21 Mar, 2017 1 commit
  32. 11 Mar, 2017 1 commit
  33. 10 Mar, 2017 1 commit
  34. 07 Mar, 2017 1 commit