On 30/10/06, Martin Sebor wrote: > > Geoffrey Winn wrote: > > On 17/10/06, Martin Sebor wrote: > > > >> > >> I think we'll need a test case to figure out what's going on. > >> (Does the SDO library execute any code at startup?) It would > >> help to reduce it as much as possible by eliminating as many > >> dependencies as you can. > >> > >> Martin > >> > > > > > > Well one thing I could do is help you set up a Tuscany SDO build > > environment. It's fairly straightforward and then you could see the > problem > > for yourself. The test program can be trivial - and I'll supply it. It > > should take about 15 mins to set it up. Let me know if that is any use. > > Sorry for the delayed response -- I just got back from a two week > trip. I expect to need to spend the next few days catching up on > email and other things but if setting up Tuscany isn't too involved > I might be able to look into it sometime in the relatively near > future (late this week or early next week). > > Martin > No problem with the response - it gave me time to do a little work on this myself. Anyway 1. Tuscany SDO should be easy to set up - and if it isn't I need to know that so I can fix it. Let me know when/if you have some time and I'll send the instructions. 2. I tried exploring this problem with a trivial hello world example, as follows. #include #include "commonj/sdo/SDO.h" using namespace std; using namespace commonj::sdo; int main() { cout << "Hello World.Start.\n"; #ifdef SDO DataFactoryPtr mdg; mdg = DataFactory::getDataFactory(); #endif cout << "Hello World.End.\n"; } If I compile without defining SDO and link with stdcxx it works fine (no surprise). If I define SDO then I have to link with the SDO library as well and that works too. This pretty strongly hints that I'm still not getting the Tuscany build modified correctly. I am building the example above with a trivial shell script, however the Tuscany build is a much more complicated collection of autoconf/automake/etc scripts. Obviously I would welcome you trying out a Tuscany build however I'm going to take another look at the way the SDO test programs are built because that seems to be the most likely suspect at the moment. Regards, Geoff.