@zach1221 can you confirm about -c option on your side ?

Ok I ran ufs_test.sh on Hercules and it looks like after @jkbk2004's changes, the rocoto_workflow.xml file skips all of the dependent tests. For Example, running with "-c" command, tests cpld_control_p8_intel, and cpld_control_p8.v2.sfc_intel run then cpld_restart_p8_intel, cpld_control_qr_p8_intel, cpld_restart_qr_p8_intel, etc are skipped, until the next non-dependency test cpld_control_ciceC_p8_intel runs. So in short, dependent tests are being skipped when creating baselines.

Ok I ran ufs_test.sh on Hercules and it looks like after @jkbk2004's changes, the rocoto_workflow.xml file skips all of the dependent tests. For Example, running with "-c" command, tests cpld_control_p8_intel, and cpld_control_p8.v2.sfc_intel run then cpld_restart_p8_intel, cpld_control_qr_p8_intel, cpld_restart_qr_p8_intel, etc are skipped, until the next non-dependency test cpld_control_ciceC_p8_intel runs. So in short, dependent tests are being skipped when creating baselines.

@DusanJovic-NOAA Could you verify that you see the same?

I see that the cpld_2threads_p8 test is now not executed when rt is running in baseline create mode (-c), which is good. But I also see that this commit 9c48b52 now specify that a lot more tests depend on their corresponding control test , which they do not depend on. I guess these 'dependency' flags are now used to skip baseline creation. Which is not correct. Like I explained, or tried to explain, in my previous comment (#2278 (comment) which for some reason is marked as 'resolved') 'baseline' and 'dependency' are two different attributes. Using just 'dependency' to describe both is not going to work in general. For example for tests that both require new baselines and depend on some other test, like:

RUN | cpld_control_gfsv17_iau                           | - noaacloud                          | baseline | cpld_control_gfsv17

or:

RUN | atm_ds2s_docn_dice                                | - noaacloud wcoss2 acorn             | baseline | cpld_control_nowave_noaero_p8

For these tests in baseline create mode (-c) baselines will not be created because they 'depend' on other test. Also adding these extra 'dependency' flag to tests that do not depend other test will unnecessarily slow down the overall execution of rt.

But, since I see that for some reason there is a pressure to merge this PR, I'm going to remove my 'request for change' which blocks the merge, but unfortunately I can not approve this PR.

@DusanJovic-NOAA thanks for the comment! hopefully, full dependency and baseline creation features can be addressed when cpld_control_gfsv17_iau is turned on next round. performance tuning as well. @ulmononian FYI: additional test requirement.

@DusanJovic-NOAA thanks for the comment! hopefully, full dependency and baseline creation features can be addressed when cpld_control_gfsv17_iau is turned on next round. performance tuning as well. @ulmononian FYI: additional test requirement.

@jkbk2004 indeed -- noted.

@BrianCurtis-NOAA can you close the request so that @FernandoAndrade-NOAA can move on for merging this pr?

@jkbk2004 I would recommend creating an issue to address Dusan's concerns on the tests that need to create baselines even if they depend on another test so you can follow up accordingly as you progress with the development of your testing framework.

Opened #2397

Ok, I'll get started with letting atmos cubed sphere to proceed with their merge.