Output precision to statistics file for time and Time Step Size is now 12

[egpuckett]

These changes are in response to issue #824 opened 3 days ago by @bangerth and subsequent email correspondence between @bangerth and @egpuckett

[bangerth]

These changes are almost certainly going to break every single one of our 310 tests.

Let's find out!
/run-tests

[egpuckett]

Oh pooh! I assume this is because the diff script is comparing digits and the changes produce an additional n digits, which causes the test to fail?

[bangerth]

Yes, that's the cause. (We do compare only up to a tolerance, but the relative difference here is larger than the tolerance we compare against :-( )

But you can download a patch file for the differences here: http://www.math.clemson.edu/~heister/aspect-logs/2b8e7463f9659ffea9e00c6b7807ca325be03beb/changes.diff . Just apply it and add the result as a second commit to your pull request.

[egpuckett]

I assume /run-tests will start automatically?

[egpuckett]

I guess so ...

[tjhei]

You also need to update the output from the PETSc tests (sorry there is no automated way to generate a patch file for that at the moment).

[egpuckett]

How do I do that? I can see the four tests that failed at

http://www.math.clemson.edu/~heister/aspect-logs/757a65550f4e59bac93405c0a90dd2f2e065e522/gccpetsc

and the log files from the tests that failed

http://www.math.clemson.edu/~heister/aspect-logs/757a65550f4e59bac93405c0a90dd2f2e065e522/gccpetsc

but, as you obviously know, the diffs in that log file are incomplete:

******* First 50 of 80 lines of diffs are:

Do I need to obtain the file

Results are stored in
/aspect-tester/build-gccpetsc/tests/output-checkpoint_02_petsc/statistics1.diff.failed

and the other three *.diff.failed files and create a patch file? If all of this is correct, then

• Can I get those files from you, or do I have to run the tests themselves here to obtain the files.
• How do I make a patch file from the *.diff.failed files? Perhaps read the Git manual, or is there a simple instructions you can send me?

I've read some documentation about creating patches in Git

http://makandracards.com/makandra/2521-git-how-to-create-and-apply-patches[](url)

However, I'm not sure I have the files I need to COMMIT from which I can generate the patch file. That's assuming I know what I'm doing, which is debatable! :-)

[tjhei]

There is no way to get access to the diffs or a patch file right now. I made a patch for you though. Can you please do

git fetch https://github.com/tjhei/aspect pr830
git merge FETCH_HEAD

[egpuckett]

Done!

[tjhei]

You broke nearly every test by changing the name of the RK2 in source/particle/integrator/rk_2.cc (from "rk2" to "RK2").

[egpuckett]

That was foolish of me. Rene warned me about this, but I failed to notice that the change was committed when I committed the change to 12 digit precision. My sincere apologies. Mea culpa ...

[bangerth]

On 04/25/2016 01:30 PM, Elbridge Gerry Puckett wrote:

That was foolish of me. Rene warned me about this, but I failed to notice that
the change was committed when I committed the change to 12 digit precision. My
sincere apologies. Mea culpa ...

Rule #1: Only one functional change per pull request or commit! :-)

[egpuckett]

Got it!