Also Jürgen and Nyall: is there anything the PSC could be funding to help you allocate the time needed to help get things to a good conclusion? PSC this is IMHO be a better use of overflow funds right now than e.g. windows CI infrastructure.
On Sun, 24 Nov 2019 at 02:38, Jürgen E. Fischer <[hidden email]> wrote:
So overall how the situation was handled doesn't seem that bad for a release exercising new major dependencies.
I'd also like to extend Tim's thanks here and publicly state my appreciation to Jürgen for an extremely difficult, thankless job which you've done SO well, SO consistently and for SO long.
(And also +1 to whoever said that this was **always** going to be a very tricky transition for the opensource geo world.It's a good thing, but thanks to proj6 the entire software stack has been *forced* to mature and adapt to modern, accurate projection handling and it's causing/caused pain all round. At least we're mostly through the worst now and it's now the users and geospatial analysts who are being forced to mature and skill up as a consequence! And all those "data scientists" who blundered around thinking spatial was just numeric x/y coordinates have an even worse learning curve inbound...!)
Moving forward, I'm thinking/wondering:
Regarding 3.10.1: there's been some work in GDAL and PROJ which may help resolve the issue which triggered this discussion. I'll test tomorrow. I also think there's potential hacky workaround we could implement if we avoid re-using existing PROJ context objects in certain places (albeit with a performance hit as a result), which I'm going to explore tomorrow as a band-aid so that we can unblock the release.
Regarding 3.4: One option we should explore is just freezing the Windows releases at 3.4.13, and leaving 3.4.13 as the final supported 3.4 release for Windows. (For reference, 3.4.16 is the planned final 3.4.x release, so we'd "lose" 3 months of supported new releases). I'd personally be +1 to this, given that the existing infrastructure cannot be easily modified to allow a proj4 based build, and adding this support is far from trivial. While it's not ideal, I think finalizing the Windows releases at 3.4.13 is much less risk and better for our users then risking projection handling issues. (For full disclosure of information I've included a list of fixes as  which are included in 3.4.14 at the time of writing which would NOT be available for Windows users as a result)
 QgsVectorFileWriter: fix axis order issue with GDAL 3 (fixes #33014) Merge pull request #33003 from rldhont/backport-32800-to-release-3_4 Fix broken QMap<QVariantList,...> finding, which causes case-insensitive comparisons to be made when resolving primary keys in the Oracle and Postgres providers [Server] Update WMS GetProjectSettings tests for round extent in GetCapabilities [Server] Test WCS Access Control: update comment [Server] Update WCS tests for round extent in GetCapabilities [Server] Update WMS tests for round extent in GetCapabilities [Server] Update WFS tests for round extent in GetCapabilities [Server] Update WMTS tests for round extent in GetCapabilities [Server] tests: Add masks [Bugfix][Server] Use floor and ceil for round extent coordinates in services capabilities [Bugfix][Server] Correctly round extent coordinates in services capabilities Wrong label in UI for "Create point for each part" on "Centroids" processing algorithm #32940 - Backport to 3.4 debian packaging: python-gdal still needed for gdal python scripts Merge pull request #32749 from qgis/backport-32743-to-release-3_4 database style manager: translatable & title case custom widgets: fix designer crash (fixes #32860) debian packaging: drop python-gdal dependency (closes #32835) [ci skip] qgsfunction: replace deprecated inspect.getargspec() to inspect.getfullargspec() oracle provider: log when ROWID is used for a missing primary key (closes #32648) also track newer CIFS translation string fix Merge pull request #32460 from alexbruy/backport-32455-to-release-3_4 osgeo4w: fix nightlies (followup 766149b1b6) In case it is needed: Update Rasterize.py for 3.4 (backport of master commit) Update CreateConstantRaster.py Merge pull request #32793 from rldhont/fix-server-wfs-add-primary-keys-to-request-for-fid-34 osgeo4w: build with proj 6 and gdal 3 Fix const [Bugfix][Server] WFS: Add primary keys to request to build Server Feature Id Fix PG 12 constraints check (provider side) Fix crash when deactivating vertex editor (fixes #32685) Fix postgis 12 adscr -> adbin consrc -> conbin "fix" gdal_calc creation options Merge pull request #32572 from qgis/backport-32527-to-release-3_4 osgeo4w: detect grass78 [RPM] Update sip references in spec file Fix copy items from value map configuration [processing] allow to select files without suffix in the Processing options dialog fix wronng paste fix SAGA Slope, Aspect, Curvature dwg/dxf import: * fix orientation of TEXT entities * also clean TEXT strings * support non-origin-based blocks * support extrusion
If i had to run this through a committee first, probably nothing of the above had happend yet and I wouldn't have time left to actually do it.
PS: Windows CI status: * appveyor: still times out (after 4 years the limit of 1h still isn't enough) * azure pipelines: runs out of disk space (limit 10GB)
PPS: I also need something to punch ;)
-- Jürgen E. Fischer norBIT GmbH Tel. +49-4931-918175-31 Dipl.-Inf. (FH) Rheinstraße 13 Fax. +49-4931-918175-50 Software Engineer D-26506 Norden https://www.norbit.de QGIS release manager (PSC) Germany IRC: jef on FreeNode _______________________________________________ QGIS-Developer mailing list [hidden email] List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer
_______________________________________________ QGIS-Developer mailing list [hidden email] List info: https://lists.osgeo.org/mailman/listinfo/qgis-developer Unsubscribe: https://lists.osgeo.org/mailman/listinfo/qgis-developer