Re: Review Request 66431: Windows: Fixed `os::read()` to use `ReadFile()`.

[Andrew Schwartzmeyer <an...@schwartzmeyer.com>]

> On April 18, 2018, 11:55 a.m., Joseph Wu wrote:
> > 3rdparty/stout/include/stout/os/windows/read.hpp
> > Lines 39-44 (patched)
> > <https://reviews.apache.org/r/66431/diff/4/?file=1994444#file1994444line41>
> >
> >     So you're saying that `ReadFile` reads all the existing data on the handle prior to returning False?    It would feel safer if we verify this with a unit test.
> 
> Andrew Schwartzmeyer wrote:
>     I think that `TEST_F(SubprocessTest, PipeOutputToFileDescriptor)` adequately ensures this is working as inspected. Let me know what you think...
> 
> Joseph Wu wrote:
>     I don't think piping output to an FD is quite what I'm thinking of.
>     
>     Perhaps a test should:
>     (1) Open a subprocess that writes to a pipe held by the test.
>     (2) The subprocess should write a substantial amount of data (more than a memory page).
>     (3) Close the subprocess before reading any data from the pipe.  Presumably this is where the broken pipe error should occur?

Done, in https://reviews.apache.org/r/66892/


- Andrew


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/66431/#review201442
-----------------------------------------------------------


On April 9, 2018, 3:53 p.m., Andrew Schwartzmeyer wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/66431/
> -----------------------------------------------------------
> 
> (Updated April 9, 2018, 3:53 p.m.)
> 
> 
> Review request for mesos, Akash Gupta, Eric Mumau, John Kordich, Joseph Wu, and Michael Park.
> 
> 
> Bugs: MESOS-8676
>     https://issues.apache.org/jira/browse/MESOS-8676
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> This can eventually support overlapped I/O.
> 
> The Windows API `ReadFile()` returns an error if the pipe is broken,
> where `_read()` did not, but this is not an error for us as the data
> is still read correctly. So we ignore it.
> 
> 
> Diffs
> -----
> 
>   3rdparty/stout/include/stout/os/read.hpp 49878e499209fa2f91fede0ebdabb8f088a9d018 
>   3rdparty/stout/include/stout/os/windows/read.hpp 8047ad590fcc46d3ec46b551472d8c518ae49cc1 
> 
> 
> Diff: https://reviews.apache.org/r/66431/diff/4/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Andrew Schwartzmeyer
> 
>