GeoTools / GeoServer PMC meeting - 2023-05-23Attending
-
Torben Barsballe
-
Jukka Rahkonen
-
Andrea Aime
Actions from prior meetings:
-
[DONE] Andrea to ask for a 2.23.1 release volunteer
Agenda
1.
GS 2.23.1 release train
2.
CSW/JDBCConfig contentious PR
3.
Backporting URL checks GSIP
4.
Planetary CRS GSIP progress
5.
H2 removal, bit by bit
Actions
-
TBD
GS 2.23.1 release train
Ian performed the releases, thank you!
Looks like announcements are the missing bit.
Release publish jobs are failing at rysnc:
https://sourceforge.net/p/forge/site-support/24659/
We might want to try replicate the publish script in the job and see if
that works by changing the rsync location (could ssh into the server and
try rsync manually, not clear how to provide credentials though.. the
release logs might contain some hints about usage of ssh-agent).
CSW/JDBCConfig contentious PR
https://github.com/geoserver/geoserver/pull/6841
Exposing CSW records of local layers against JDBCConfig is slow. Need other
PSC members concerned with GeoServer architecture to chime into the
discussion.
Backporting URL checks GSIP
https://github.com/geoserver/geoserver/wiki/GSIP-218
Landed on main. Discussion ongoing about backports to stable. Remote checks
enabled by default or not?
Torben: disabled by default, very clear message in the release notes that
they should be updating. Too many people assume patch version updates are
safe, it might make people stop updating in fear of production breakage.
Jukka: +1
Planetary CRS GSIP progress
Some changes landed in GeoTools, have a gt-iau-wkt authority: 2000 extra
codes, it’s going to be a plugin. Can configure a shapefile and a geotiff
and do map preview (that needs a GeoServer PR
<https://github.com/geoserver/geoserver/pull/6833>, just getting it merged).
Also added support for authority in property file datastore, which allows
to write tests in GeoServer. Working now on capabilities document
generation.
GeoTools has no idea about the planet it’s dealing with , PROJ seems to
have extended the database:
data/sql/iau.sql:INSERT INTO ellipsoid VALUES('IAU_2015',49900,'Mars (2015)
-
Sphere',NULL,'IAU_2015',499,3396190.000000,'EPSG','9001',NULL,3396190.000000,0);
but also:
data/sql/esri.sql:INSERT INTO celestial_body VALUES('ESRI', 'Mars', 'Mars',
3393400.0);
Using this proj denies reprojection between different planetary bodies, for
example:
cs2cs "IAU:49900" "EPSG:4326"
proj_create_operations: Source and target ellipsoid do not belong to the
same celestial body
From proj.db
semi_major_axis
ESRI Atlas Atlas 16000.0
ESRI Helene Helene 16000.0
https://en.wikipedia.org/wiki/List_of_Solar_System_objects_by_size
So no, we cannot just compare by diameter.
Fun situation, OCG API Features mandates output in WGS84… what if your data
is for Mars!!
Example WKT2 for Jupiter and Mars:
https://github.com/OSGeo/PROJ-CRS-Explorer/blob/master/wkt2/IAU_2015/59900.txt
https://github.com/OSGeo/PROJ-CRS-Explorer/blob/master/wkt2/IAU_2015/49900.txt
See also Mars prime meridian:
https://mars.nasa.gov/gallery/atlas/PIA03207.html
H2 removal, bit by bit
Replacing GWC disk quota and KML superoverlay databases with HSQLDB. Enough
to have GeoServer core free of H2 dependencies.
Already used for EPSG database, works fine for non spatial database, does
not support spatial.
GWC pull request ready and GS incoming. This will also allow H2GIS to
become an official extension (will be incompatible with some other
extensions, but at least usable).
In time we hope to upgrade other parts like NetCDF, GeoFence, WPS and
importer async status storage.