I have pushed a fix to the barry/snes-monitor-vi-residual and next branch.
Note that it is ok for DMDestroyVI() to be called "extra" times. Barry > On Mar 18, 2015, at 4:35 PM, Ed Bueler <[email protected]> wrote: > > Barry and Jed -- > > Actually, I think I do see a bug that has caused me some problems, namely > that when RSLS diverges, stuff is corrupt. This seems to be because > DMDestroyVI() was not called when the diverged cause a "break" out of the for > loop in SNESSolve_VINEWTONRSLS(). Part of the problem was the indent level. > ;-) > > I've decided to follow this advice: > """ > *For very small changes you can clone the petsc repository, create a new > branch on your local repository from master, apply and commit your changes > there, and use git format-patch origin/master to create the patch. Send the > patch to [email protected]. > """ > > The attached should explain itself, but note I branched off of > barry/snes-monitor-vi-residual, and thus did > git format-patch barry/snes-monitor-vi-residual > to generate the patch. > > Is this an easy to apply form, compared to me being confused about my fork? > > Ed > > -- > Ed Bueler > Dept of Math and Stat and Geophysical Institute > University of Alaska Fairbanks > Fairbanks, AK 99775-6660 > 301C Chapman and 410D Elvey > 907 474-7693 and 907 474-7199 (fax 907 474-5394) > <0001-DMDestroyVI-needs-to-go-after-end-of-for-loop-or-els.patch>
