[jira] [Commented] (MESOS-1000) Clang build broken on 0.18.0 master

["Dominic Hamon (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/MESOS-1000?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13900450#comment-13900450 ] 

Dominic Hamon commented on MESOS-1000:
--------------------------------------

2. We could also consider replacing uses of unreachable with an invariant check, and always return valid objects. Ie,
  if (validCondition)
    return validObject();
  return UNREACHABLE();

becomes

  CHECK(validCondition);
  return validObject();

This is clearer and more readable.

> Clang build broken on 0.18.0 master
> -----------------------------------
>
>                 Key: MESOS-1000
>                 URL: https://issues.apache.org/jira/browse/MESOS-1000
>             Project: Mesos
>          Issue Type: Bug
>          Components: libprocess
>    Affects Versions: 0.18.0
>         Environment: Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn)
>            Reporter: Till Toenshoff
>              Labels: clang, stout
>
> There are at least three new issues preventing mesos to build on clang.
> 1. ../../src/slave/containerizer/mesos_containerizer.hpp:174:8: error: 'mesos::internal::slave::MesosContainerizerProcess::exited' hides overloaded virtual function [-Werror,-Woverloaded-virtual]
> That one is rather easy to overcome; simply add something like the following to that header;
>   // Prevent clang warning on hidden overloaded virtual function.
>   using process::ProcessBase::exited;
> 2. ../../src/slave/containerizer/launcher.cpp:120:12: error: conversion from 'Unreachable' to 'Try<pid_t>' is ambiguous
> Can overcome by a nasty but simple disambiguation of the returned value by updating that line towards:
>   static_cast<Try<int> >(UNREACHABLE());
> In this case, I guess it would be better to revisit the Unreachable implementation.
> 3. ../../src/slave/containerizer/mesos_containerizer.cpp:220:18: error: no matching function for call to 'defer'
> For that one, I am still trying to find out which overload of defer would be the best match and hence should possibly be updated accordingly.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)