Hi, On Thu, 2009-06-18 at 11:36 +0200, Andreas Halm wrote: > Hi, > > And first of all thanks for your answers. > > > Gerrit gave the details, but just to make it clear: we *strongly* warn > > against > > copy-n-pasting to generate your FCs. Use the .fcd mechanism! That's is > > a much, > > much easier way, as there are just too many side dependencies that you > > do not > > want to mess with (like the list above), especially if things need to > > be > > extended or changed on the inside. > > > > We do not give any guarantees that the internal structure of FC does > > not change. > > We do guarantee that the fcds you write now will continue to work > > I understand that, however i am under tight time constrains and I felt it > might be easier to first go the copy-and-paste-way,
ok, where did you copy from (e.g. from which OpenSG class) ? > and later "repair" the code. I already did the same thing with the include > directories, > as I have no idea how cmake works and I already had working > solution/projects. So I > added all the OpenSG include paths using a custom property sheet and decided > to move to cmake when I have time for it. ok. Short question, how do you compile OpenSG ?. Weird crashes can easily be a mismatch of compile time parameters, especially crashes that early. With the risk of getting another that should have been documented, you can easily integrate custom libs into the main OpenSG cmake build structure (OSG_EXTRA_EXTERNAL_MODULES). I just committed an example that uses it and which should be easily adapted : Examples/CustomIntegration/AlgorithmStage/GPUVolRT/AsAlgorithmStage > Unfortunately now I am at a point where I am not sure I can continue that > way. My solution compiles finally, but the generated application crashes > instantly on startup. It looks like the WinMain is about to initialize > global and static variables, and in the process of initializing the > FieldContainerType variable of one of my custom node types it can't copy the > xml string variable (which I don't use anyway, I left it at the default > value ""). could you send a stack trace ?. Is that really 'one of' or more 'the first' ? One easy thing that is always worth a try is to use "foo" as the value instead of "". > Maybe I did something wrong when implementing the new nodes. I might have to > create the fcd files earlier than I thought. By the way, is there a tutorial > about that? If not, add it to my wish list ;-) Now that is something where cut&paste makes sense ;). The documentation wish is noted though. > > > - What the hell happened to beginEditCP/endEditCP > > > > Gone for good (yippiie! ;). Just forget about them. All you do need to > > do is > > call commitChanges() (or commitChangesAndClear()) after you're done > > creating a > > block of objects for your graph (or, a little more concretely, when you > > switch > > from writing to reading from FCs). > > I already read about commitChanges(), but I thought there must be something > else I have to do. I guess when creating nodes using the fcd mechanism > everything needed is automatically generated. > what exactly is not working ? The crash mentioned above should have nothing to do with it. kind regards, gerrit ------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensing option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ Opensg-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/opensg-users
