GeoTools / GeoServer PMC meeting - 2024-03-26Attending
-
Andrea Aime
-
Peter Smythe
-
Jukka Rahkonnen
-
Jody Garnett
Actions from prior meetings:
-
Andrea: make a GSIP to amend the module graduation rules
-
Next meeting:
-
Interesting tile seeding speedup PR
-
Sponsorship update from Jody
-
jody: send invite to geoserver-security email list
-
jody to set up an rst-fix branch to collect fixes, ideally with
automation of mkdocs_translate
Agenda
-
GSIP 222 and 223 status
-
Tile creation speedup PR
-
mkdocs update
-
NetCDF dependency upgrade and backwards incompatibility
-
Chit chat
Actions
-
GSIP 222 and 223 status
GSIP 222, Graduating the Raster Attribute Table community module:
-
Approved, past 10 days with enough +0 and +1, no negatives
-
Andrea to make a PR to graduate the module
GSIP 223, Community module graduation, amending generality rule
-
Approved, past 10 days with enough +0 and +1, no negatives
-
Andrea to make a documentation PR to update the procedures
Tile creation speedup PR
Bunch of PRs at various levels to speed up tile generation:
-
https://github.com/geosolutions-it/imageio-ext/pull/299 (sub-image
encoding)
-
https://github.com/GeoWebCache/geowebcache/pull/1227 (improved
concurrency)
-
https://github.com/geoserver/geoserver/pull/7457
-
Enables the imageio-ext speedup
-
Encode first tile on the request thread, delegate the others to a
secondary thread pool
-
Poor interaction with the mass-seed case, being worked on
Good stuff, thank you Mitchell Bösecke!
It has been a while since I've heard from him, we might have to complete
the development.
Problem with JMH tests. It should not be part of the quality test, because
the GitHub actions/jenkins is not reliable. For reference, GDAL is also
running performance tests, with 30% tolerance due to the unreliable
environment. See mail thread
https://www.mail-archive.com/gdal-dev@anonymised.com/msg39829.html
How to handle it:
-
@Ignore and let them be run interactively
-
Move them to the online profile?
-
Make yet another profile?
-
Add alongside other excludes:
https://github.com/search?q=repo%3Ageoserver%2Fgeoserver+onlinetest+language%3A"Maven+POM"&type=code&l=Maven+POM
Only two use cases in a large code base, likely to get forgotten…
mkdocs update
Collect the rst changes.
Idea:
-
Start from https://github.com/geoserver/geoserver/pull/7462
-
Remove the mkdocs
-
Collect the RSTs only
-
Share the load of fixing the RSTs across various people
-
Backport the changes
-
Actually run the mkdocs
-
https://github.com/jodygarnett/translate
Here are the instructions for running translate script:
1.
https://jodygarnett.github.io/translate/translate/migrate/
mkdocs_translate init
mkdocs_translate scan
mkdocs_translate migrate
2.
Common problems:
1.
Too much indenting (results in a block quote)
2.
Stray ```` results in the rest of the file being wrong
3.
Examples also from previous RST PR:
https://github.com/geoserver/geoserver/pull/7429/files#diff-7802f6ce973902c35df3d9c6670250c7191549c206ffbcccf20b6ae2582bb573L121
NetCDF dependency upgrade and backwards incompatibility
I am sorry Steve, it does look like this is an actual compatibility break
that cannot be addressed.
-
GRIB file configured in GeoServer 2.25 will not actually work in future
GeoServer 2.26!
-
they changed 30-40 mapping tables in the library
-
name of the variable is constructed at runtime (not read from the file
it is constructed from header and how time is understood)
-
Take down the layers and start from scratch; updates notes have
instructions.
Can we provide an option in the upgrade guide to just hand-edit the files
instead? Only for single files…
Extra details and examples here:
https://osgeo-org.atlassian.net/browse/GEOT-7547
Chit chat
CVE Disclosure feedback / articles showing up.
-
Difference between those seeking exposure vs to address an issue