Another thought. In your error handler you call MPI_Abort. I think this is causing problem for my application. In my application I use both real and complex versions of PETSc, so I initialize MPI myself, then initialize PETSC, then initialize libmesh.
Can I still have libmesh throwing exceptions (because it is useful to do checks), but not catch them inside the libmesh? Michael. On 08/26/2015 06:53 PM, John Peterson wrote: > > > On Wed, Aug 26, 2015 at 4:23 PM, Michael Povolotskyi > <mpovo...@purdue.edu <mailto:mpovo...@purdue.edu>> wrote: > > What happens is as follows: > I use the gdb editor. > > > Ah, so it's a debugging issue, not a runtime issue... > > > I run the code there until the exception is thrown. Since it is > not caught inside the code, the code stops and I use gdb command > "up" and "print" for debugging. > Usually my stack depth is less than 10, so this is easy. > > > And this only works when exceptions are *disabled* in libmesh? > Because this is about how it works for me when exceptions are enabled too. > > > Now, when your handler is called the depth has been increased up > to 100 levels and somehow I got into destructors of my objects. > Doing "up" command many times I was able to get to my exception > throwing place, so I could debug, but this was inconvenient. > May be there is a problem in my code, I do not know. > > > And this is an *uncaught* exception, correct? I suppose it's possible > that your implementation actually unwinds the stack for an uncaught > exception (I have never seen an implementation that does this) and > that's why you are seeing all the extra frames... > > By the way, you can always type "frame n" in gdb to go directly to the > frame you want, avoiding typing "up" dozens of times, > https://sourceware.org/gdb/onlinedocs/gdb/Selection.html > > -- > John ------------------------------------------------------------------------------ _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
