[Lldb-commits] [lldb] r224310 - Instead of rolling our own, use the C++11 sanctioned solution

[Zachary Turner]

Heh, interesting.  Didn't know that.  I will change to #include and check
it in, thanks for the explanation of why it worked.

On Tue Dec 16 2014 at 11:10:20 AM Enrico Granata <egranata at apple.com> wrote:

> It says #import because of my silly brain, actually
> And, yes, it compiles on OS X - clang is quite happy to oblige ObjC
> directives even if it’s not ObjC code
>
> On Dec 16, 2014, at 11:03 AM, Zachary Turner <zturner at google.com> wrote:
>
> Why does this say #import instead of #include?  Does this actually compile
> on MacOSX?  I'm assuming it's just a typo for #include, but I'm curious why
> this ever worked at all on any platform.
>
> On Mon Dec 15 2014 at 8:34:14 PM Enrico Granata <egranata at apple.com>
> wrote:
>
>> Author: enrico
>> Date: Mon Dec 15 20:34:13 2014
>> New Revision: 224310
>>
>> URL: http://llvm.org/viewvc/llvm-project?rev=224310&view=rev
>> Log:
>> Instead of rolling our own, use the C++11 sanctioned solution
>>
>> Modified:
>>     lldb/trunk/source/Core/ConstString.cpp
>>
>> Modified: lldb/trunk/source/Core/ConstString.cpp
>> URL: http://llvm.org/viewvc/llvm-project/lldb/trunk/source/
>> Core/ConstString.cpp?rev=224310&r1=224309&r2=224310&view=diff
>> ============================================================
>> ==================
>> --- lldb/trunk/source/Core/ConstString.cpp (original)
>> +++ lldb/trunk/source/Core/ConstString.cpp Mon Dec 15 20:34:13 2014
>> @@ -11,6 +11,8 @@
>>  #include "lldb/Host/Mutex.h"
>>  #include "llvm/ADT/StringMap.h"
>>
>> +#import <mutex>
>> +
>>  using namespace lldb_private;
>>
>>
>> @@ -184,25 +186,16 @@ protected:
>>  // we can't guarantee that some objects won't get destroyed after the
>>  // global destructor chain is run, and trying to make sure no destructors
>>  // touch ConstStrings is difficult.  So we leak the pool instead.
>> -//
>> -// FIXME: If we are going to keep it this way we should come up with some
>> -// abstraction to "pthread_once" so we don't have to check the pointer
>> -// every time.
>>  //----------------------------------------------------------------------
>>  static Pool &
>>  StringPool()
>>  {
>> -    static Mutex g_pool_initialization_mutex;
>> +    static std::once_flag g_pool_initialization_flag;
>>      static Pool *g_string_pool = NULL;
>>
>> -    if (g_string_pool == NULL)
>> -    {
>> -        Mutex::Locker initialization_locker(g_pool_
>> initialization_mutex);
>> -        if (g_string_pool == NULL)
>> -        {
>> -            g_string_pool = new Pool();
>> -        }
>> -    }
>> +    std::call_once(g_pool_initialization_flag, [] () {
>> +        g_string_pool = new Pool();
>> +    });
>>
>>      return *g_string_pool;
>>  }
>>
>>
>> _______________________________________________
>> lldb-commits mailing list
>> lldb-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
>>
>
> Thanks,
> *- Enrico*
> 📩 egranata@.com ☎️ 27683
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20141216/a394c5cc/attachment.html>