Wt

Hello,

Can I deduce from your comment that all is working well in the Release configuration?

In debug, you need to link to wtd.lib, wthttpd.lib (notice the 'd' in the end of the library) instead of wt.lib, wthttp.lib etc. These libraries are compiled against the debug versions of the C runtime libraries.

Best regards,

Wim.

Unfortunately, that isn't a fair assessment.. oh and I should have mentioned that I am linking against the correct versions of Wt, and did a complete rebuild of the project and working directories to ensure that I am linking and loading the correct libs & binaries.

Neither the release nor the debug work correctly. Though it is a little difficult to determine if the issues seen in release and debug are the same.

I'm not exactly sure how I might be certain they are, can you suggest anything?

If you are interested in the problem, please see the attached call stack from which the error is thrown..

Hi Wim,

I can confirm after some more testing that the issue is only present when running the debug build.

Cheers,

Pat

And also this for _SCL_SECURE discussion:

http://redmine.webtoolkit.eu/boards/1/topics/1048

It would be useful to know your compiler version and a sinppet of the offending code. If you're getting a "Vector Iterators Incompatible" exception then it's probably not an _SCL_SECURE build issue as that would either be a weird crash or a link error depending on your compiler version. It's more likely that you are comparing iterators from different collections, such as is shown in this stack overflow post:

http://stackoverflow.com/questions/8421623/vector-iterators-incompatible

Also be aware that vector iterators become invalid if the vector reallocates its buffer, so any iterator created before you do, say, a push_back() may become invalid. This is especially true of the iterator returned by end(). Again on StackOVerflow:

http://stackoverflow.com/questions/15855634/why-am-i-getting-vector-iterators-incompatible

Chances are that you are misusing the STL in your code somewhere. If that's the case then you can thank Microsoft for the excellent iterator checking! I haven't looked at the call stack you posted, but the site that the exception is thrown from (make sure you set up your debugger to stop on exceptions being thrown) should pinpoint where you've gone wrong.

Cheers,

Dan

Hello,

The stack trace points to some Wt code. This is the place of where the invalid compare is happening according to the stack trace:

template<typename T>
inline bool erase(std::vector<T>& v, const T& value)
{
  typename std::vector<T>::iterator i = std::find(v.begin(), v.end(), value);

  if (i != v.end()) {  // Offending line, #115
    v.erase(i);
    return true;
  } else
    return false;
}

I'm really confused about how this can happen (other than the reason I already specified, or as consequence of an ordinary memory error). Is this a Wt example you're trying to compile?

BR,

Wim.

Hey Wim, Dan,

Thanks for the feedback.

Firstly Wim - No it's not an example, it's part of an application that I am developing. At this stage it is rather messy and hard to isolate different sections of the code which interact with the web server, but this is what I am working on to pin point the problem.

Dan - Setup up is as follows

• I am using the pre-compiled 3.3.4 RC2 binaries of Wt, distributed with boost 1.56
• The application itself is compiled using MFC in a shared DLL (there is a desktop part of the application which provides console like function)

anyone, I had a sneaking suspicion that the problem was to do with client sessions timing out. In which case, I assumed that the destruction of the client object would be initiated, whilst in parallel, the server was trying to post updates to this client and causing my problem. But after the session timeout has elapsed, the object is not being destroyed.. What is the expected behaviour following a session timeout?

I think the problem has been found... Another thread was calling WSAStartup multiples times, along with WSACleanup.. AFAIK, this only needs to be called once per thread.. After fixing this, we don't seem to be having the same problem.. I'll leave the code running for a little while longer before I say this issue is fixed..

On another note, why does the creation of a WApplication instance throw an exception?

Normally WApplication does not throw an exception. What exception is thrown?

BR,

Wim.

Hi Wim,

Apologies for the delay.

Using the pre-built binaries for vc 10 for Wt 3.3.4 (stable release).

I compiled and ran the hello world example provided. Turning the debugger on to catch all the exceptions thrown, the attached call stack precedes the exception thrown during creation of the client application.

The exception does not create a crash, so it could potentially be a Visual Studio debugger problem? (Though I have also tested in vc 13, and obtain the same problem).

BR,

Pat

Hello Pat,

I tested the prebuilt msvs 2013 64bit build with MSVS2013 update 4. When I turn on all exceptions, I see a callstack that resembles yours, i.e. the stack is:

    KernelBase.dll!RaiseException()    Unknown
    [External Code] 
>   wtd.dll!Wt::WEnvironment::getCookie(const std::basic_string<char,std::char_traits<char>,std::allocator<char> > & cookieName) Line 435   C++
    wtd.dll!Wt::WebSession::handleRequest(Wt::WebSession::Handler & handler) Line 1388  C++
    wtd.dll!Wt::WebController::handleRequest(Wt::WebRequest * request) Line 713 C++

That is an expected exception, and is caught in WebSession.C. Note that if you enable 'break on exception', the debugger will break when an exception is thrown, irrespective of the fact if it was expected or not, so your program may break on properly handled exceptions.

I don't see however the incompatible iterators exception. Do you get it on MSVS 2013?

Wim.