Loki-transform: Remove custom entry point options

[mlange05]

Note: This sits on top of PR #464 and requires CLOUDSC CLOUDSC PR 106 to be merged first.

Note: For testing only at this stage.

Following the conversion to config-file based pipelines, this PR now removes all the custom entry point options under loki-transform.py convert and their associated use in CMake macros. It also removes the now unused loki_transform_convert entry point.

[github-actions]

Documentation for this branch can be viewed at https://sites.ecmwf.int/docs/loki/465/index.html

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.03%. Comparing base (304921f) to head (7dfefc4).
Report is 11 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #465   +/-   ##
=======================================
  Coverage   96.03%   96.03%           
=======================================
  Files         226      226           
  Lines       40581    40587    +6     
=======================================
+ Hits        38972    38978    +6     
  Misses       1609     1609           

Flag	Coverage Δ
lint_rules	96.39% <ø> (ø)
loki	96.02% <ø> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mlange05]

This should now be ready, but still relies on some downstream dev branches. I'll put this into Review mode, once these are merged and I've removed the respective re-pointing commits.

[reuterbal]

This is very pleasing to see!

In addition to removing the processing of the CMake options we should directly remove them from the corresponding function interfaces - this will make sure to hard fail with an error instead of silently ignoring deprecated options, which I think is better at this stage.

[reuterbal]

With these removed, we should now also remove the corresponding arguments from the CMake function - that will make sure that any options that have been specified are not silently ignored but do actually raise an error in CMake directly. For that, ll. 226-231 need to look like this:

    set( options NO_PLAN_SOURCEDIR COPY_UNMODIFIED CPP CPP_PLAN )
    set( single_value_args TARGET COMMAND MODE FRONTEND CONFIG PLAN )
    set( multi_value_args SOURCES HEADERS DEFINITIONS INCLUDES )

Please also update the docstring to match the reduced interface.

[reuterbal]

With these removed, the argument definition to loki_transform (ll. 48-57 in loki_transform.cmake) should be updated accordingly:

    set( options CPP )
    set( oneValueArgs COMMAND MODE FRONTEND CONFIG BUILDDIR )
    set( multiValueArgs OUTPUT DEPENDS SOURCES HEADERS INCLUDES DEFINITIONS OMNI_INCLUDE XMOD )

Please also update the docstring of loki_transform accordingly.

Also, since we only have the CPP option left now, it might be worthwhile to just remove this macro and instead inline the remaining 3 lines of code into loki_transform and loki_transform_plan directly

[reuterbal]

Might be helpful to mention the specified mode here, to spot typos quickly:

Suggested change

	msg = '[Loki] ERROR: Pipeline or transformation mode not found in config file.\n'
	msg = f'[Loki] ERROR: Pipeline or transformation mode "{mode}" not found in config file.\n'

[reuterbal]

Pet peeve, I find this much more readable:

Suggested change

	if not mode in config.pipelines:
	if mode not in config.pipelines:

[mlange05]

@reuterbal Indeed, very satisfying! 😉 Thanks for the pointers, I think it's ready for another look.

[reuterbal]

Many thanks!