[Vortex] Linking Problem (SASL/TLS prebuilt binaries + VC2005)

[Maxim Reality]

Hi,

Found another probem: I get an access violation error when calling
vortex_sasl_set_plain_validation().



BR,

Juhani

ps. Is this list working actually?


On Fri, Feb 20, 2009 at 11:58 PM, Maxim Reality <itsmekissme at gmail.com> wrote:
> Hi,
>
> I did some further research using dumpbin and found out that symbol
> names were mangled in the obj file calling the functions. Adding
> extern "C" around Vortex headers
> fixed the problem but is not really that pretty solution. Could we fix
> this into the official headers?
>
> Any thoughts on the callback issue btw?
>
> BR,
>
> Juhani
>
>
> On Fri, Feb 20, 2009 at 9:26 PM, Maxim Reality <itsmekissme at gmail.com> wrote:
>> Hi,
>>
>> I am now trying to change my code to use new 1.1 version of Vortex,
>> but encountered some problems with TLS and SASL support. I get some
>> unresolved symbols when linking against prebuilt libvortex-1.1.lib,
>> libaxl.lib
>> libvortex-tls-1.1.lib and libvortex-sasl-1.1.lib libraries.
>>
>> "error LNK2019: unresolved external symbol "int __cdecl
>> vortex_sasl_accept_negotiation(struct _VortexCtx *,char const *)"
>> (?vortex_sasl_accept_negotiation@@YAHPAU_VortexCtx@@PBD at Z) referenced
>> in ..."
>>
>> Any ideas on that? Should I try to build the Vortex source myself?
>> It's strange that the main vortex lib seems to be ok, though.
>>
>> Another problem I'm having is that the two callbacks for finding
>> certificates and private key file locations do not have any user data
>> field (axlPointer user_data) defined and this causes me trouble since
>> I am using 'user_data' to store "this" pointer of a class for proper
>> encapsulation (redirecting functions callbacks to methods). I have a
>> Vortex context as a class member and thus need a this pointer to
>> access it. This was not a problem earlier, since there were no concept
>> of separate contexts before. I could kludge this by using a static
>> variable, but would prefer doing it the right way (so I guess what I'm
>> asking is that whether we could change the callback signature to
>> contain user_data as a last parameter). What do you think?
>>
>> To sum up what I have done so far in order to get Vortex 1.0 code to
>> compile as 1.1 is:
>>
>> 1. Changed negociate in some function names to negotiate.
>> 2. Changed initialization to use vortex context
>> 3. Added context as a first parameter for all function calls.
>> 4. Had to make a kludge to provide the correct context for certificate
>> and private key file locations as no user_data is provided with the
>> callbacks.
>> 5. Added #include statements for vortex TLS and SASL headers.
>> 6. Changed to use Vortex 1.1 lib and aded 1.1 TLS and SASL libs as well
>> 7. The code compiles now fine but get unresolved linker errors.
>>
>>
>> Thanks in advance for any help,
>>
>>
>> Juhani
>>
>