Permalink
Commits on Mar 27, 2016
  1. @vito
Commits on Aug 28, 2015
  1. @vito @totherme

    dialer can be injected

    this allows the caller to decide things like TCP keepalive
    configuration, timeouts, etc. etc.
    
    [#101647542] [#101732194]
    vito committed with totherme Aug 24, 2015
Commits on Feb 25, 2015
  1. @vito @xoebus

    reduce log level of property handlers

    Signed-off-by: Chris Brown <cbrown@pivotal.io>
    vito committed with xoebus Feb 25, 2015
Commits on Feb 3, 2015
  1. @vito @atulkc

    bump to go 1.4.1 [#86810456]

    Signed-off-by: Atul Kshirsagar <atul.kshirsagar@gmail.com>
    vito committed with atulkc Feb 3, 2015
Commits on Oct 31, 2014
  1. @vito @maxbrunsfeld

    Add FakeClient

    Signed-off-by: Max Brunsfeld <mbrunsfeld@pivotal.io>
    vito committed with maxbrunsfeld Oct 31, 2014
Commits on Oct 22, 2014
  1. @vito @johannespetzold

    Lookup: return ErrContainerNotFound explicitly

    this is more convenient for clients to check against
    
    Signed-off-by: Johannes Petzold <jpetzold@pivotal.io>
    vito committed with johannespetzold Oct 22, 2014
Commits on Oct 21, 2014
  1. @vito @sykesm

    add get/set/remove property client/server impl

    we have expressly ignored the tests under client/ in favor of covering
    it all under server/ by using the client.
    
    [#79618100]
    
    Signed-off-by: Matthew Sykes <matthew.sykes@gmail.com>
    vito committed with sykesm Oct 21, 2014
Commits on Sep 12, 2014
  1. @vito

    one more go

    vito committed Sep 12, 2014
Commits on Aug 28, 2014
  1. @vito

    flush stdout/stderr after process exits

    this fixes a race with quickly printing-and-exiting processes
    vito committed Aug 28, 2014
  2. @vito @Zteve

    when failing to destroy, clear destroying lock

    [finishes #77768828]
    
    Signed-off-by: Steve Powell <spowell@pivotal.io>
    vito committed with Zteve Aug 28, 2014
Commits on Aug 12, 2014
  1. @vito

    regenerate fakes

    vito committed Aug 12, 2014
Commits on Aug 11, 2014
  1. @vito @stewartnickolas

    prevent concurrent container destruction

    [finishes #76616270]
    
    Signed-off-by: Stew Nickolas <nickolas@us.ibm.com>
    vito committed with stewartnickolas Aug 11, 2014
Commits on Aug 8, 2014
  1. @vito
Commits on Jul 31, 2014
  1. @vito @sykesm

    add server/request logging

    Signed-off-by: Matthew Sykes <matthew.sykes@gmail.com>
    vito committed with sykesm Jul 31, 2014
  2. @vito
  3. @vito

    stream out: close backend stream when conn. breaks

    also introduce very early work on lager
    vito committed Jul 30, 2014
Commits on Jul 22, 2014
  1. @vito

    Merge pull request #16 from cloudfoundry-incubator/tty-in-run

    tty can be configured as part of process spec
    vito committed Jul 22, 2014
Commits on Jul 19, 2014
  1. @vito

    tty can be configured as part of process spec

    this allows configuration like window size to be configured immediately;
    otherwise there is a race with the process starting, where sending window size
    may not be effective
    vito committed Jul 19, 2014
Commits on Jul 14, 2014
  1. @vito

    Merge pull request #15 from cloudfoundry-incubator/winsize

    add window size to process protocol
    vito committed Jul 14, 2014
Commits on Jul 12, 2014
  1. @vito

    add window size to process protocol

    this can be sent by the client to resize the tty attached to the process.
    
    there is no response sent back; the message is of low importance and may be
    sent many times (when a user is literally resizing their terminal), so it's
    probably not worth the roundtrip.
    vito committed Jul 12, 2014
Commits on Jul 11, 2014
  1. @vito

    regenerate fakes (no-op)

    Signed-off-by: Sheel Choksi <sheelc@pivotallabs.com>
    vito committed with Sheel Choksi Jul 11, 2014
  2. @vito
Commits on Jul 9, 2014
  1. @vito

    Stdout/Stderr are no longer io.Closers

    this actually ended up being inconvenient in many places. a very common use
    case is emitting logs over a persistent connection; having that connection
    close after the process exits is pretty inconvenient. it also assumes you have
    an io.Closer in the first place.
    
    this also brings it yet closer to *exec.Cmd
    vito committed Jul 9, 2014
  2. @vito
Commits on Jul 8, 2014
  1. @vito

    remove extraneous TODO

    vito committed Jul 8, 2014
Commits on Jul 7, 2014
  1. @vito

    server passes stdin payloads to backend

    closes the stream when a Stdin payload is received with no data
    vito committed Jul 7, 2014
  2. @vito
  3. @vito

    rework Run/Attach interface

    * stdin/stdout/stderr are done via io.Readers/io.WriteClosers, rather than
      everything going over one channel
    * Env is now a simple []string, just to be a bit more like exec.Cmd, which
      also tries to be a OS-agnostic interface, despite K=V being 'Unixy'
    * the attach/run API calls are now hijacked HTTP connections, so that stdin
      may be sent across the wire later
    * rather than consuming the full stream until a payload with an ExitStatus is
      detected, the caller can call process.Wait() to block until it exits, which
      will return you its exit status.
    * an 'error' field has been added to ProcessPayload to signify .Wait() failing
      server-side. this will bubble up to the client, which will return an error
      on its respective .Wait().
    vito committed Jul 5, 2014
Commits on Jul 4, 2014
  1. @vito

    generate connection interface

    vito committed Jul 4, 2014
  2. @vito
  3. @vito
Commits on Jul 1, 2014
  1. @vito
Commits on Jun 23, 2014
  1. @vito
  2. @vito

    fix fake client

    vito committed Jun 23, 2014
  3. @vito

    remove now-unused drain package

    vito committed Jun 23, 2014