Hi Robert, On Thu, Dec 18, 2008 at 10:51 AM, Robert Osfield <[email protected]> wrote: > Hi Mattias, > > I can't think of any reason why multi-threaded would be any different > than single thread when running with a static build. I'll do a static > build here and see if I see any issue when multi-threading. I think > to resolve this type of problem we'll need to roll out testing to the > wider community to see what a range of machine, hardware, drivers and > OS's think of the static build.
ok. I may be setting up a dell xps laptop over the holidays. If so I'll try a static build on it. > > Do you have to make any mods to the OSG to get things running with static > build? no. no mods . Built fine. Mattias > > Robert. > > On Thu, Dec 18, 2008 at 12:30 AM, Mattias Helsing <[email protected]> wrote: >> Hi Robert, >> >> A clean build went through earlier today (including wrappers). This >> time with vc90. The results were the same - I just get the cow when >> running osgstaticviewer singlethreaded. I haven't had time to >> investigate why the multithreaded modes don't work, I haven't even had >> time to up the osg_notify_level and see what's reported there. >> >> The results I get from running with multithreaded modes is "broken >> mirrors" (lines and points everywhere). Hardware and drivers are the >> same as my previous static build with vc80 so could still be driver >> related. Is there anything you would like me to try or some special >> information you'd like me to produce regarding the static build? >> >> cheers >> Mattias >> >> On Mon, Dec 15, 2008 at 8:21 PM, Robert Osfield >> <[email protected]> wrote: >>> HI Mattias, >>> >>> On Mon, Dec 15, 2008 at 5:57 PM, Mattias Helsing <[email protected]> >>> wrote: >>>> The results I got were a bit unclear so I was not sure what to report :-) >>>> >>>> On the good side - I can view the cow.osg with the staticviewer. On >>>> the bad - only with --SingleThreaded. >>>> >>>> After a rebuild I started the viewer with cow.osg and got "broken >>>> mirrors". I'm not using the latest nvidia drivers so don't know is >>>> these are at play. Running single threaded works fine. >>>> >>>> After this I checked my visual studio projects to make sure my /OPT >>>> setting was there and it wasn't. So im thinking that it was the >>>> rebuild that fixed the problem. I need to start a fresh rebuild to be >>>> sure. I can't start that until tomorrow. Get back as soon as I can >>> >>> OK. Not sure what to make of the results so far... look forward to see >>> how you get on with a frash build. >>> >>>> On a related thought. How should we handle the applications in static >>>> builds. Using the current setup with static plugins, either osgviewer >>>> nor osgconv will be very interesting. Should they be excluded from the >>>> build? Looking at it from the other end one might argue that the >>>> plugins should always be shared objects (or they are not really >>>> plugins). Asking stupid questions is one of my strengths but flame me >>>> nicely ;-) >>> >>> osgviewer, osgconv and osgfilecache and osgarchive are general purpose >>> apps that rely on the plugins to provide the bulk of their abilities. >>> osgversion would be OK though. I'm just tweak the >>> applications/CMakeLists.txt to reflect this and will check in once >>> I've done a clean static build. >>> >>> In the examples directory only the osgstaticviewer and the >>> examples/CMakeLists.txt already limits the static build to just this >>> app so thats OK as it is. >>> >>> In the case of plugins, if you make the app static then you really >>> need to make the plugins static as there is potential collision about >>> what version of the OSG the plugins pull in vs what the application >>> pulls in. There is also the issue of each plugin need it's own static >>> linking of the core OSG libraries so sizes would go up through the >>> roof. For these two reasons I would have though that static core libs >>> + dynamic libs won't be sensible, if at all possible. >>> >>> Robert. >>> >>> >>> Robert. >>> _______________________________________________ >>> osg-submissions mailing list >>> [email protected] >>> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org >>> >> _______________________________________________ >> osg-submissions mailing list >> [email protected] >> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org >> > _______________________________________________ > osg-submissions mailing list > [email protected] > http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org > _______________________________________________ osg-submissions mailing list [email protected] http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
