bring back the old value of ini_pseudo_initial_date

[mandresm]

Hi @JanStreffing,

I think you were the one chaging ini_pseudo_initial_date value from prev_run.general.start_date to initial_date. This PR reverts that change.

The problem with the change you made is that people like @mathanase trying to run a continuation of a experiment need to manually specify ini_pseudo_initial_date. If they don't then the restart links have incorrect negative time in their names because initial_date is used. With the old approach, this people get automatically the correct ini_pseudo_initial_date for simulations that continue from the last date into the future.

For simulations that need to continue with those restarts but from a different date (branching of a control run finished in 2850 and starting a historical in 1850), the user would still need to specify ini_pseudo_initial_date manually. We could think of a way of actually shorting that problem as well automatically.

Note that ini_pseudo_initial_date is only used for branchoff restarts.

[nwieters]

Hi,
I also tried a branchoff experiment with AWICM3. Therefore I tried out the changes in this PR. In general it worked for me for esm_runscript.
But then esm_master comp-awicm3-develop was not running but gave the following error:

Traceback (most recent call last):
  File "/work/ba1264/a270089/conda/envs/esm_tools_macv2sp/bin/esm_master", line 33, in <module>
    sys.exit(load_entry_point('esm-tools', 'console_scripts', 'esm_master')())
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_master/cli.py", line 113, in main
    main_flow(parsed_args, target)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_master/esm_master.py", line 57, in main_flow
    complete_setup = SimulationSetup(user_config=user_config)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_runscripts/sim_objects.py", line 113, in __init__
    self.config = prepare.run_job(self.config)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_runscripts/prepare.py", line 20, in run_job
    helpers.evaluate(config, "prepare", "prepare_recipe")
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_runscripts/helpers.py", line 71, in evaluate
    config = esm_plugin_manager.work_through_recipe(
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_plugin_manager/esm_plugin_manager.py", line 159, in work_through_recipe
    config = getattr(submodule, workitem)(config)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_runscripts/prepare.py", line 647, in finalize_config
    config.finalize()
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 3305, in finalize
    self.run_recursive_functions(self)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 3331, in run_recursive_functions
    recursive_run_function([], config, "atomic", marked_date_to_date_object, config)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/provenance.py", line 1004, in inner
    output = func(tree, rhs, *args, **kwargs)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 2040, in recursive_run_function
    right[key] = recursive_run_function(
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/provenance.py", line 1004, in inner
    output = func(tree, rhs, *args, **kwargs)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 2040, in recursive_run_function
    right[key] = recursive_run_function(
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/provenance.py", line 1004, in inner
    output = func(tree, rhs, *args, **kwargs)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 2004, in recursive_run_function
    right = func(tree + [None], right, *args, **kwargs)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/provenance.py", line 1004, in inner
    output = func(tree, rhs, *args, **kwargs)
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_parser/esm_parser.py", line 2643, in marked_date_to_date_object
    entry = Date(actual_date, config["general"]["calendar"])
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_calendar/esm_calendar.py", line 265, in __init__
    self._init_from_str(indate, calendar=Calendar())
  File "/work/ab0995/a270089/esm_tools_macv2sp_dars/src/esm_calendar/esm_calendar.py", line 287, in _init_from_str
    date, time = indate2.split("_")
ValueError: too many values to unpack (expected 2)

I tried this with esm_tools, version 6.55.0

What of course workes is setting ini_pseudo_initial_date: "${prev_run.general.start_date}" in my runscript.yaml.

[mandresm]

I think this can be merge once the tests pass or are approved by me, right @JanStreffing? @FinnHeu did test the feature already in #1405

[mandresm]

Hi, I also tried a branchoff experiment with AWICM3. Therefore I tried out the changes in this PR. In general it worked for me for esm_runscript. But then esm_master comp-awicm3-develop was not running but gave the following error:

...

I tried this with esm_tools, version 6.55.0

What of course workes is setting ini_pseudo_initial_date: "${prev_run.general.start_date}" in my runscript.yaml.

Hi @nwieters, I could reproduce this issue and I think I have now solved it.

[mandresm]

This PR needs to wait for #1329 to be merged to avoid conflicts on the tests (normally not needed but there are some special circumstances around #1329 that result into it blocking other PRs)

[mandresm]

Closes #1405

[mandresm]

#bump