Raised by @hannorein and @mschwamb on #495: rename the flag column to
error or success so code reads if (!error).
Two things to settle before anyone implements it.
They are not interchangeable. 0 means converged today, so if (!error)
works with the values unchanged, but if (success) is false for every good fit
unless the values also invert. "error or success style" covers both.
Renaming is a breaking change. "flag" appears 58 times across
orbitfit.py, unpack.py, convert.py and 16 test files, and 7 fixture CSVs
in tests/data/ carry it as a column header. Output already written stops
loading. Cheap before v1.0, not after.
Alternative: keep flag, add a derived boolean column (accepted). Reads
the same, breaks nothing, inverts nothing, and gives -sf (#490) a named
predicate. Also avoids labelling -1 (not attempted) and 8 (bookkeeping) as
errors, which they are not.
Flag values are documented in #495.
Raised by @hannorein and @mschwamb on #495: rename the
flagcolumn toerrororsuccessso code readsif (!error).Two things to settle before anyone implements it.
They are not interchangeable. 0 means converged today, so
if (!error)works with the values unchanged, but
if (success)is false for every good fitunless the values also invert. "error or success style" covers both.
Renaming is a breaking change.
"flag"appears 58 times acrossorbitfit.py,unpack.py,convert.pyand 16 test files, and 7 fixture CSVsin
tests/data/carry it as a column header. Output already written stopsloading. Cheap before v1.0, not after.
Alternative: keep
flag, add a derived boolean column (accepted). Readsthe same, breaks nothing, inverts nothing, and gives
-sf(#490) a namedpredicate. Also avoids labelling -1 (not attempted) and 8 (bookkeeping) as
errors, which they are not.
Flag values are documented in #495.