In your last mail you denoted that the problem is caused by the new 64bit version of localtime.Gerhard Gappmeier wrote:Ragarding VS2005. I know that problem from other projects. * time_t is 64 bit now: Use the define _USE_32BIT_TIME_T to make it backward compatibleWell, the crash is a problem in the "compilers code" and not in our code - I don't know if setting this will fix our problem here.
So switching back to 32 bit may fix the problem. Just an assumption.
cool, thxThe fuzz tests seems to be running smooth now - *I've checked in your dissector in SVN 21760* :-)
You're right. ua_ prefix is used for the general protocol layer code which is written by hand.Two questions after I had a look at your code: - some files use the prefix ua_ while others use opcua_ (I would prefer opcua_ for all such source files)
opcua_ prefix is used for generated datatype parsers.
But this is not consistent also, because the opcua_simpletypes.c is also written by hand.
I've no problem with changing all to opcua_.
Numeric is used as return value for the function parseServiceNodeId in ua_application_layer.c. Retrieving this Id is the main purpose- in some files (opcua_simpletypes.c, ua_application_layer.c, ua_security_layer.c) you use tvb_get_... to explicitly get some data without actually using it (e.g. Numeric and NSId in ua_application_layer.c) Is there a reason for this (e.g. future use), or is this just left over from development / debugging and should be fixed?
of this function.
NsId is not used at the moment. I just used this for debugging.
The UA services will always use NsIdx zero.
in opcua_simpletypes.c (parseNodeId) both the numeric value and NsIdx are not used.
We can remove both.
I'll send you a patch these days.