Ah. Well, my general sentiment remains. Warnings should be an *aid*, not a *demand*. If they’re too annoying, it isn’t hard to define rules to filter out most such warnings.
Bill On Apr 8, 2015, at 10:28 AM, Jed Brown <[email protected]> wrote: > William Gropp <[email protected]> writes: >> The issue is whether you want to be a slave to some compilers buggy warning >> code. > > Sure, but this is not a compiler bug. The statement really is > unreachable when PETSC_USE_ERRORCHECKING (default), but it can't be > fixed without breaking a different configuration. It could be fixed in > this case with more macros, but the problem is more general and we > probably need to turn off some warnings.
