[raknet] Re: 2.32 released

  • From: Kevin Jenkins <rakkar@xxxxxxxxxxxxxx>
  • To: raknet@xxxxxxxxxxxxx
  • Date: Tue, 03 May 2005 23:48:16 -0700

Sorry for the late reply.  My mail server was down.

Source/Rand.cpp is capitalized.

I'm not sure what happened to RakNetGetTime.  I think Dalfy changed that 
when he made the DOxygen port unless I just forgot.  I guess just use 
RakNet::GetTime from now on.

Xavier wrote:

> Two issues with 2.33 (and I am guessing others):
> 
> 1) Source/Rand.cpp has (ought) to be capitalized (Linux case-sensitivity 
> issues). This got un-capitalized sometime since 2.28 (which is when I 
> build my automake script files); Windows didn't care, but when I 
> unpacked the source on Windows and checked it in, it confused SVN and 
> the update on the Linux build server lost track of Rand.cpp.
> 
> 2) AutopatcherTest.cpp calls RakNetGetTime() without any params (which 
> should be correct) but MSVC 7.1 (stock VS.NET 2k3 install here) chokes 
> on it; even tho the GetTime() method is declared (void), the macro 
> processor is putting something in there that causes a link failure. 
> Taking out the (void) and replacing it with emptiness in the GetTime.h 
> and GetTime.cpp files fixes that (for me anyway).
> 
> FYI
> Greg
> 
> Xavier wrote:
> 
>>Kevin
>>
>>Is there any sort of patching procedure for RakNet? In lieu of one...
>>
>>AutoPatcher.cpp in Source/AutoPatcher needs to have <stdlib.h> included
>>as well, for the "system" call, on Linux (it's defined by default on
>>Windows). It's harmless on Windows, so it doesn't need to be bracketed
>>with #ifdef.
>>
>>Greg
>>
>>Kevin Jenkins wrote:
>>
>>
>>>2.32 is released.
>>>
>>>For Windows, I replaced Sleep with WaitForSingleObject on a timer.  In 
>>>theory this should greatly improve both performance and consistency. 
>>>Please let me know if you find this to be the case.
>>>
>>>For reasons I don't understand, it wouldn't compile CreateWaitableTimer 
>>>and SetWaitableTimer until I globally defined _WIN32_WINNT=0x0410 .  I 
>>>shouldn't have to do this because it should already be done from 
>>>windows.h - but it wasn't.  If anyone knows of a resolution to this 
>>>issue please let me know.  In the meantime, if you use the source 
>>>directly you will also need to define that in your project settings if 
>>>you want to use WaitForSingleObject.  Otherwise, it will revert to using 
>>>Sleep.  There is a #pragma message that indicates which you are using 
>>>when you build.  The pre-built lib and dll uses WaitForSingleObject.
>>>
>>>Linux continues to use Sleep.  If anyone knows the Linux equivalent 
>>>please let me know (or better yet - the code).
>>>
>>>This version also fixes a variety of problems with setting the password.
>>>
>>>As a side note, I attempted to fix IO completion ports but due to 
>>>structural changes I wasn't able to get it to work within a reasonable 
>>>amount of effort.  In a future version I will re-evaulate the changes 
>>>needed to get it to work again as well as investigate if it offers any 
>>>performance benefit over the current optimized system.
>>>
>>>
>>
>>
>>
>>
> 
> 
> 
> 

Other related posts: