On Fri, 4 May 2012, Robert wrote: > How? I ran the sample with (1.0, 0, 0) again in debug mode and got no > additional warning or assertion.
This is quite surprising - the code ought to have been hitting the diff.size() > tolerance test around line 1000 of fe_map.C. It does for me, and although you're using a slightly different configuration I can't see anything (other than compiler flags omitting -DDEBUG) that would cause those warnings to be missed. > Log of the run with translation=(2.0,0,0): > >> for mode in dp pd d p; do ./bc_bug-dbg $mode; echo >> "-------------------------"; done > Adding Dirichlet BC > Adding Periodic BC > Constraints before init: > Local DoF Constraints: > Constraints after init: > Local DoF Constraints: > Constraints for DoF 1: (0,1) rhs: 0 > Constraints for DoF 2: (3,1) rhs: 0 > Constraints for DoF 4: rhs: -0.1 > Constraints for DoF 5: rhs: 0 > Constraints for DoF 6: rhs: 0 > Constraints for DoF 7: rhs: -0.1 This bug, on the other hand, I could reproduce. And thank you very much for finding it! The broken code wasn't just a bad interaction between Dirichlet and periodic constraints, it was a bad interaction between Dirichlet and *any other* constraints - people doing adaptivity alongside Dirichlet boundaries might have had (much more subtle) errors as well. Would you test your app with r5610 and see that my fix corrects the problems there too? Copying this to libmesh-devel to ask: does anyone have any experimental or in-flux code in the SVN trunk right now? I need to decide whether to tag trunk as 0.7.4 or just backport this one fix to 0.7.3.2. --- Roy ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Libmesh-devel mailing list Libmesh-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-devel
