I have indeed tested the fix - it's very easy to check for as we have all kinds of 3D problems. I just flipped the switch to turn on adaptivity on a few of these problems and checked the solutions (and sidesets) before and after the patch was applied. I think Derek is still amazed that it slipped under the radar for so long. He kept asking me how we've managed to not get messed up by this for so long. We have had problems with some of out 3D adaptive problems though which is how we found the bug in the first place!

I looked through the other elements and didn't see any others that were labeled with different orders of vertices but it's possible that some of the higher order elements might have the same bug. Going through that embedded matrix though for the higher order elements is non-trivial so you may want to double check if you have time. Just out of curiosity, why are the child elements in the hex class out of order in the first place relative to the vertices?

Hope you are feeling better,
Cody


On Jan 28, 2011 2:14pm, Roy Stogner <roystgnr@ices.utexas.edu> wrote:
>
>
> Nice job, thanks!  Sorry I've been silent (sick) this week, but your
>
> fix in r4179 looks pretty much ideal.
>
>
>
> This was a very serious bug, so I'd like to put out a 0.7.0.4 release
>
> with the fix in.  A few questions first:
>
>
>
> Is it safe to assume you've tested this fix?  ;-)
>
>
>
> Does anyone know about any regressions since 0.7.0.3 that need to be
>
> looked at before I tag/tarball a new release?
>
> ---
>
> Roy
>