More early resolve issues #414
Conversation
…ipts_regression tests
…ipts_regression tests
Revert "Revert "More early resolve issues"" This reverts commit fe0ae4d.
| @@ -424,7 +424,8 @@ | |||
|
|
|||
| <batch_system MACH="yellowstone" type="lsf" version="9.1"> | |||
| <queues> | |||
| <queue walltimemax="24:00" jobmin="1" jobmax="8">caldera</queue> | |||
| <queue walltimemax="24:00" jobmin="1" jobmax="1" jobname="case.lt_archive">hpss</queue> | |||
There was a problem hiding this comment.
Add a jobname attribute to the queue node. This allows me to force the case.lt_archive script to use the hpss queue on yellowstone.
|
Does this fix #406 or just address part of it? |
| @@ -330,6 +330,8 @@ def get_submit_args(self, case, job): | |||
| if name is None: | |||
| submitargs+=" %s"%flag | |||
| else: | |||
| if name.startswith("$"): | |||
There was a problem hiding this comment.
In config_batch.xml the submit_args node references other xml variables, we need to strip the $ from the front of the variable so that we can resolve it. We have to do this explicitly because this is an attribute not a node,
|
I believe that it fixes 406 |
|
This all looks fine. I could swear some of these exact same changes were merged yesterday though, so that confuses me. |
|
They were merged Monday - but I found several failing tests related to that perl build-namelist so I reverted the merge and fixed those issues before issuing the PR again. Sorry for the confusion. |
|
Ah, forgot about that, thanks. |
This change unresolves some variables in the case that were being resolved too early, this resulted in some of the perl code breaking which I've put a bandaid on in the hope that it will be replaced soon.
Test suite: scripts_regression_tests
Test baseline:
Test namelist changes:
Test status: bit for bit
Fixes #406
User interface changes?:
Code review: