Hi David,
I reviewed you changes, but rather than adopt these they have spurred
me on to a wider refactor of how the RequestQueue/ReadQueue are
managed, and have centralised more of the functionality into
RequestQueue and replaced the sort() method usage with an O(n)
algorithm for pruning and
Hi,
Wang Rui writes:
The '#pragma pack' command may not work for all platforms. Maybe we
could add a '#ifdef' statement to handle this.
GCC is also compatible with those pragma directives, see
http://gcc.gnu.org/onlinedocs/gcc/Structure_002dPacking-Pragmas.html
. Other platforms should ignore
Hi Wang Rui,
Sorry for the slow reply. Lack of expertise in the area of VS #pragma
makes it awkward for me to provide any authoritative view, so I'm in a
listening and learning mode :-)
I've just reviewed your changes and I feel that the use of push and
pop isn't something that is prone to
Hi Jim,
On Tue, Mar 23, 2010 at 6:19 PM, Jim Brooks jimbl...@gmail.com wrote:
Something else that is confusing CMake? Perhaps a Cmake bug? Which
version of CMake are you using?
What is happening is that BSD splits POSIX pthreads into
standard and non-standard headers:
-r--r--r-- 1 root
A small update, when I add the required
INCLUDE (CheckPrototypeExists)
To allow the use of CHECK_PROTOTYPE_EXISTS I get the CMake error:
CMake Error at src/OpenThreads/pthreads/CMakeLists.txt:3 (INCLUDE):
include could not find load file:
CheckPrototypeExists
CMake Error at
Hi Jim,
On Wed, Mar 24, 2010 at 4:59 PM, Robert Osfield
robert.osfi...@gmail.com wrote:
The other possible solution might be to use the compilation of local
source to see if the method is there. I will now look into this...
I have just tested out the following:
I have just tested out the following:
CHECK_FUNCTION_EXISTS(pthread_setaffinity_np HAVE_PTHREAD_SETAFFINITY_NP)
IF(HAVE_PTHREAD_SETAFFINITY_NP)
# double check that pthread_setaffinity_np is available as FreeBSD
header doesn't contain required function
CHECK_CXX_SOURCE_COMPILES(
#include
On 20/03/10 21:42 , Ulrich Hertlein wrote:
W.r.t SCOPE local variable definition, I believe this will lead to
warnings about an unused variable when the OSG is built with
notification disabled. Perhaps a local #define would do the trick on
providing the local SCOPE. Personally though I'm not
