ert
ert copied to clipboard
equinor
Simplify EnsembleConfig
please hit me up when you start on this, i'm currently trying to write tests for ensemble config, and I'm removing a bit of unneeded code in the process.
to summarize a bit what has been done so far - we moved grid and refcase (files) from ecl config to ensemble config, where it's actually used.
in the course of that work, we saw that we did not have new config parser style tests for ensemble config, and started working on creating those, in particular extending the config dict generator to create options and option arguments for ensemble config.
we covered some keywords, but not all. the keywords that were covered, are
-
GEN_DATA -
FIELD
the uncovered ones that I can see are
-
GEN_KW -
SURFACE -
SUMMARY -
SCHEDULE_PREDICTION_FILE
with their respective arguments, which can be found (possibly partly) in the docs, and in the config dict code path of ensemble config, and of course in the C code. so the config dict generator should be extended to cover the remaining keywords.
what also happened was that for the covered keywords, we investigated which of the options were actually used, and discovered that some were unused, and removed them (and code associated to them).
finally, one might also look at possibly existing old tests to draw inspiration on what the options and arguments and their respective possible values are (we have not done this).
and last but not least, no work has been done yet to port C code to python, besides the mentioned removal of unsed code paths.
instead of TODOs in the code, I'm gonna note two things here. In config dict generator, we should
- also cover the key
ConfigKeys.GEN_KW_TAG_FORMAT - we should investigate if
REPORT_STEPSas an argument toGEN_DATAshould be sorted (it's a list of ints, somehow used for reporting perhaps, and supposed to match something, cf. #4072 )
gonna reopen this, as we want to have another go at lifting moar parts of ensemble config to python