Accidentely replied privately only, reposting to the ml. On Wed, Mar 02, 2011 at 10:59:16AM +0100, jeremy rosen wrote: > > Since threading in general is tricky, I wonder how easy will it be > > to > > accidently break OMP due to seemingly unrelated changes? > > that's a good question, I'm not too worried for the animation part > because it is well contained... > I think OMP will only be used on a few well-contained area and not > everywhere throughout the code, which should help quite a bit.
Ok. > > Would it make > > sense to document what code is depending on OMP and which > > preconditions > > it relies upon? > > I'm not really sure what to document What expectations OMP makes. For example it expects a variable to be read only during its execution, but if somebody changes that to be read-write it will break OMP. > and how to make sure people read that documentation... That might be hard, but if there's no documentation it will be impossible. (Realistically I fear it won't be read until something breaks.) > Usually it boils down to protecting a variable which means adding a > #pragma omp criticall(varname) around all calls to that variable. So > i'm not too afraid of people missing that... Well as long as the OMP parts are in `well-contained areas' the problem may not occur and when it does it should be easy to spot. My fear is when larger parts are converted and suddenly break to a seemingly unrelated change. -- Regards, Mark de Wever aka Mordante/SkeletonCrew _______________________________________________ Wesnoth-dev mailing list Wesnoth-dev@gna.org https://mail.gna.org/listinfo/wesnoth-dev
