search

idaholab / moose

Multiphysics Object Oriented Simulation Environment
https://www.mooseframework.org
GNU Lesser General Public License v2.1
1.7k stars 1.04k forks source link

Add test harness test to run an arbitrary command #7410

Closed dschwen closed 8 years ago

dschwen commented 8 years ago

Description of the enhancement or error report

Woooo, scary! But the ability to run an arbitrary command (or at least another python script in the same directory) would allow us to generate data files on the fly (for example big EBSD data files with thousands of lines could be generated with a few lines of python). Using prereq those scripts could be launched before the tests that need their output data.

Rationale for the enhancement or information for reproducing the error

Simplify automatic test data generation.

Identified impact

Fewer commits of large datafiles.

permcody commented 8 years ago

We already have this capability. Take a look On Wed, Jul 27, 2016 at 4:39 PM Daniel Schwen notifications@github.com wrote:

Description of the enhancement or error report

Woooo, scary! But the ability to run an arbitrary command (or at least another python script in the same directory) would allow us to generate data files on the fly (for example big EBSD data files with thousands of lines could be generated with a few lines of python). Using prereq those scripts could be launched before the tests that need their output data. Rationale for the enhancement or information for reproducing the error

Simplify automatic test data generation. Identified impact

Fewer commits of large datafiles.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/idaholab/moose/issues/7410, or mute the thread https://github.com/notifications/unsubscribe-auth/AC5XIPQyuRG7Lr5-cnzynFGGetvr_SCdks5qZ94SgaJpZM4JWsyJ .

dschwen commented 8 years ago

Ugh, great... I guess

dschwen commented 8 years ago

What's that tester called?

permcody commented 8 years ago

It's not a tester. It's built into the RunApp Tester (I think) Ben put it in for something similar so I don't know off hand. On Wed, Jul 27, 2016 at 5:39 PM Daniel Schwen notifications@github.com wrote:

Reopened #7410 https://github.com/idaholab/moose/issues/7410.

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub https://github.com/idaholab/moose/issues/7410#event-737190862, or mute the thread https://github.com/notifications/unsubscribe-auth/AC5XIE1kke6YuDlLqWrdxBIDtrnaTVvUks5qZ-xIgaJpZM4JWsyJ .

dschwen commented 8 years ago

post_command? That is not a great design. Why not have a separate command tester. Then you can chain it however you want (pre or post) using prereq.

dschwen commented 8 years ago

Why did @bwspenc even add this. It is used in neither bison nor grizzly.

bwspenc commented 8 years ago

@dschwen, as explained in issue #2573, we added it to do some postprocessing on our assessment cases automatically after we ran them, but as you noted, we haven't gotten around to adding that to any of our tests, and it's been a couple of years now. That's a little embarrassing, but I think it's still a useful option. That sounds a little different from what you want anyway.

permcody commented 8 years ago

Perhaps we can replace post_command with a different design. On Wed, Jul 27, 2016 at 6:08 PM Ben Spencer notifications@github.com wrote:

@dschwen https://github.com/dschwen, as explained in issue #2573 https://github.com/idaholab/moose/issues/2573, we added it to do some postprocessing on our assessment cases automatically after we ran them, but as you noted, we haven't gotten around to adding that to any of our tests, and it's been a couple of years now. That's a little embarrassing, but I think it's still a useful option. That sounds a little different from what you want anyway.

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub https://github.com/idaholab/moose/issues/7410#issuecomment-235759599, or mute the thread https://github.com/notifications/unsubscribe-auth/AC5XIMqmh1BRK_wYr-leHMvsx1ISOQw3ks5qZ_MLgaJpZM4JWsyJ .

bwspenc commented 8 years ago

Sure, I guess when this tester that runs another app gets implemented, we can just chain it. We could remove post_command and convert that test of it over to use this new tester. That's fine with me.

dschwen commented 8 years ago

Yeah, that's the plan. Separate tester that can be chained using prereq. That way I can also run it before a test.

On Wed, Jul 27, 2016, 6:25 PM Ben Spencer notifications@github.com wrote:

Sure, I guess when this tester that runs another app gets implemented, we can just chain it. We could remove post_command and convert that test of it over to use this new tester. That's fine with me.

— You are receiving this because you were mentioned.

Reply to this email directly, view it on GitHub https://github.com/idaholab/moose/issues/7410#issuecomment-235762338, or mute the thread https://github.com/notifications/unsubscribe-auth/AAMWPiFKXo5gGCm3AXF08ycbdQfkEoWCks5qZ_cDgaJpZM4JWsyJ .