I added some initializations in CVS to silence those warnings, but they are rather meaningless, in the sense that it is just a case of VC's code analyzer not being able to prove that if a cosmic ray were to strike the CPU in just the right manner and a bit was to flip, the value might be used unitialized. Technically those cases can never be reached, although the WSASendTo case was just using 0 where I needed a NULL (too much C++ on the brain). Oh well, sometimes it is easier to just do what the compiler says to make it stop nagging. :)
On 12/29/2011 05:57 PM, Ruud van Gaal wrote:
Having just read a blog from John Carmack on static code analysis tools, I decided to start analyzing my own code. (see http://altdevblogaday.com/2011/12/24/static-code-analysis/ ) Part of it is ENet, quite the latest version. These are the 3 results: 1>d:\source\trunk\dev\src\libs\enet\peer.cpp(743) : warning C6001: Using uninitialized memory 'reliableSequenceNumber': Lines: 715, 717, 718, 719, 720, 721, 723, 726, 739, 741, 742, 743 1>d:\source\trunk\dev\src\libs\enet\protocol.cpp(222) : warning C6001: Using uninitialized memory 'outgoingCommand': Lines: 173, 174, 175, 176, 178, 179, 189, 210, 222 1>d:\source\trunk\dev\src\libs\enet\win32.cpp(244) : warning C6387: 'argument 6' might be '0': this does not adhere to the specification for the function 'WSASendTo': Lines: 224, 225, 227, 244 Not sure how the first 2 might be problems; perhaps someone else has a better clue? Cheers, Ruud van Gaal _______________________________________________ ENet-discuss mailing list [email protected] http://lists.cubik.org/mailman/listinfo/enet-discuss
_______________________________________________ ENet-discuss mailing list [email protected] http://lists.cubik.org/mailman/listinfo/enet-discuss
