Repository navigation
Writing masked columns with serialize_method="data_mask" has no effect #18704
Description
Description
Writing a table to ecsv, I should be able to have masked sky coordinates have a separate mask column, but this does not work.
Expected behavior
The mask is written as a separate column.
How to Reproduce
# Put your Python code snippet here.
from astropy.table import QTable
from astropy.coordinates import SkyCoord
from astropy.units import Unit as U
from astropy.utils.masked import Masked
qt = QTable()
qt["sc"] = SkyCoord(ra=Masked([1.0, 2.0, 3.0], [True, False, False]),
dec=Masked([5.0, 6.0, 7.0], [True, False, False]),
unit="hourangle,deg")
qt["q"] = Masked([0.0, 8.0, 9.0], mask=[False, True, False]) << U("km")
qt.write("test.ecsv", serialize_method="data_mask", overwrite=True)This gives
cat test.ecsv
# %ECSV 1.0
# ---
# datatype:
# - {name: sc.ra, unit: deg, datatype: float64}
# - {name: sc.dec, unit: deg, datatype: float64}
# - {name: q, unit: km, datatype: float64}
# - {name: q.mask, datatype: bool}
# meta: !!omap
# - __serialized_columns__:
# q:
# __class__: astropy.units.quantity.Quantity
# unit: !astropy.units.Unit {unit: km}
# value: !astropy.table.SerializedColumn
# __class__: astropy.utils.masked.core.MaskedNDArray
# data: !astropy.table.SerializedColumn {name: q}
# mask: !astropy.table.SerializedColumn {name: q.mask}
# sc:
# __class__: astropy.coordinates.sky_coordinate.SkyCoord
# dec: !astropy.table.SerializedColumn
# __class__: astropy.coordinates.angles.core.Latitude
# unit: &id001 !astropy.units.Unit {unit: deg}
# value: !astropy.table.SerializedColumn
# __class__: astropy.utils.masked.core.MaskedNDArray
# data: !astropy.table.SerializedColumn {name: sc.dec}
# frame: icrs
# ra: !astropy.table.SerializedColumn
# __class__: astropy.coordinates.angles.core.Longitude
# unit: *id001
# value: !astropy.table.SerializedColumn
# __class__: astropy.utils.masked.core.MaskedNDArray
# data: !astropy.table.SerializedColumn {name: sc.ra}
# wrap_angle: !astropy.coordinates.Angle
# unit: *id001
# value: 360.0
# representation_type: spherical
# schema: astropy-2.0
sc.ra sc.dec q q.mask
"" "" 0.0 False
29.999999999999996 6.0 8.0 True
44.99999999999999 7.0 9.0 False
Versions
import astropy
astropy.system_info()# Paste the result here
Activity
pllim commentedon Jul 20, 2026
Someone proposed #20094 but it might be AI-generated/assisted. Is it worth enabling CI for?
nomad3 commentedon Jul 28, 2026
Yes, I believe it is worth enabling full CI. To clarify, this is an AI-assisted contribution: I used OpenAI Codex and Anthropic Claude Code as development and review assistants, with human direction and review throughout. I have reviewed and understand the submitted implementation and tests, take responsibility for the contribution and its licensing compliance, and will engage directly with reviewer questions and feedback.
The branch has been rebased onto current
mainwithout changing either patch. Local validation passed the table, ECSV, and masked-table suites (2,930 passed, 183 skipped, 14 xfailed), the FITS and available HDF5 connector tests (175 passed, 123 skipped), and the full pre-commit suite on the changed files. I added the disclosure and current validation to the PR body and marked the PR ready for review. I would appreciate full CI being enabled when convenient.