Closed dasmy closed 9 years ago
@aeslaughter Can you comment on whether this is acceptable? If so... @dasmy would you mind submitting a Pull Request for this?
Thanks!
I think that this type of numbering would be fine. @dasmy if you create a pull request I will take a closer look and fix up the error that you are still getting.
Also, could you add a simple test for the desired behavior. For example, utilize simple_diffusion.i but enable sequence=true
and set output_on='initial timestep_end nonlinear'
. This should produce a whole series of files. Thanks.
OK, I will prepare the appropriate PR including a testcase tonight or tomorrow.
Am 22/01/15 um 17:29 schrieb Andrew E Slaughter:
I think that this type of numbering would be fine. @dasmy https://github.com/dasmy if you create a pull request I will take a closer look and fix up the error that you are still getting.
Also, could you add a simple test for the desired behavior. For example, utilize simple_diffusion.i but enable |sequence=true| and set |output_on='initial timestep_end nonlinear'|. This should produce a whole series of files. Thanks.
— Reply to this email directly or view it on GitHub https://github.com/idaholab/moose/issues/4573#issuecomment-71049176.
Dr. Mathias Winkel Centre for Computational Medicine in Cardiology (CCMC)
University of Lugano Institute of Computational Science Via Giuseppe Buffi 13 CH-6900 Lugano
Tel: +41 (0)58 666 4973 Email: mathias.winkel@usi.ch Web: http://icsweb.inf.unisi.ch/cms/index.php/people/141-mathias-winkel.html
Thanks! On Thu, Jan 22, 2015 at 1:47 PM Mathias Winkel notifications@github.com wrote:
OK, I will prepare the appropriate PR including a testcase tonight or tomorrow.
Am 22/01/15 um 17:29 schrieb Andrew E Slaughter:
I think that this type of numbering would be fine. @dasmy https://github.com/dasmy if you create a pull request I will take a closer look and fix up the error that you are still getting.
Also, could you add a simple test for the desired behavior. For example, utilize simple_diffusion.i but enable |sequence=true| and set |output_on='initial timestep_end nonlinear'|. This should produce a whole series of files. Thanks.
— Reply to this email directly or view it on GitHub https://github.com/idaholab/moose/issues/4573#issuecomment-71049176.
Dr. Mathias Winkel Centre for Computational Medicine in Cardiology (CCMC)
University of Lugano Institute of Computational Science Via Giuseppe Buffi 13 CH-6900 Lugano
Tel: +41 (0)58 666 4973 Email: mathias.winkel@usi.ch Web: http://icsweb.inf.unisi.ch/cms/index.php/people/141-mathias-winkel.html
— Reply to this email directly or view it on GitHub https://github.com/idaholab/moose/issues/4573#issuecomment-71075082.
Apparently the output is performed (can be seen with
sequence=true
), but all produced files/steps contain identical physical time information. This might be formally correct but disallows ParaView from showing them individually.This only seems to apply to Steady Executioner because then
dt==0
.A possible patch might read (I did not run the testsuite, yet)
This works well when
sequence=false
because then the timesteps are assigned different pseudo-times. Thus, the intermediate steps of the nonlinear solver can be visualised in ParaView as a time series even for steady solvers.However, if
sequence=true
, it still does not work. I assume the reason for this is the fact thatexodus_num
is reset to 1 whenever a new Exodus file is started (seeExodus::outputSetup()
, line 133 inExodus.C
). Accordingly, from the visualisation point all different files still belong to the same timestep. I tried to avoid resettingexodus_num
, but this breaks the exodus output: It fails withDo you maybe have an idea on fixing this issue?
I could also live with the current state of the patch (i.e. it works if
sequential=false
and does not become worse ifsequential=true
and thus is an improvement). If we agree on some approach, I can prepare a pull request for this.