[SunBlack (SunBlack)]

• Changes comply with the maintainer guide.
• SHA512s are updated for each updated download.
• The "supports" clause reflects platforms that may be fixed by this new version, or no changes were necessary.
• Any fixed CI baseline and CI feature baseline entries are removed from that file, or no entries needed to be changed.
• All patch files in the port are applied and succeed.
• The version database is fixed by rerunning ./vcpkg x-add-version --all and committing the result.
• Exactly one version is added in each modified versions file.

mapbox-polylabel removed the C++ code in mapbox/polylabel#124. As mapnik requires it, they created a fork of it (see mapnik/mapnik#4567), therefore replace the port.

We could also move mapnik-polylabel into the source code of mapnik, similar to mapnik-vector-tile

[Billy O'Neal (BillyONeal)]

GPT 5.6 Sol reports:

The Mapnik 4.3.0 port does not control the new default-enabled combined input plugins. The port still passes obsolete feature options in portfile.cmake, while upstream now uses USE_PLUGIN_INPUT_GDAL_OGR, USE_PLUGIN_INPUT_POSTGIS_PGRASTER, and USE_PLUGIN_INPUT_TILES (dispatcher, defaults).

Consequently, default builds require undeclared dependencies and fail while locating GDAL across twelve tested Windows, Linux, and Android triplets. The PostGIS/PGRaster and tiles plugins similarly introduce undeclared PostgreSQL, SQLite, Boost, and platform-specific requirements (PostGIS/PGRaster, tiles).

[...]

Blocking finding: Mapnik's new plugin options are uncontrolled

The vcpkg port still maps the input-gdal, input-ogr, input-postgis, and input-pgraster features to the old separate options in ports/mapnik/portfile.cmake. In Mapnik 4.3.0, the plugin dispatcher no longer uses those options. Instead, it selects combined plugins through USE_PLUGIN_INPUT_GDAL_OGR and USE_PLUGIN_INPUT_POSTGIS_PGRASTER, and also selects the new tiles plugin through USE_PLUGIN_INPUT_TILES; see upstream plugins/input/CMakeLists.txt. All three are default-enabled in upstream CMakeLists.txt, but the port passes none of them.

I have not dug into the guts of the failures here because builds are broken. Drafting due to build failures, I am not making a concrete statement that all of the above need be addressed to merge.

[Kai Pastor (dg0yt)]

I suggest we keep this combined in input-gdal. I believe that wanting one without the other will be very unlikely. (It could still be achieved via triplet settings.)

[SunBlack (SunBlack)]

That was my first thought, too. But, as you've seen yourself, there's another OGR feature further down. So for now, I've decided to stay with the existing structure. I think we could look into merging some things there when the PR is ready, since the CI time for this port is really quite long. Right now, though, the PR is failing anyway because I still don't have any idea what the issue is with the Linux CI, and when I give the log to AI, the solution it suggests seems a bit strange to me (even though I haven't tested it yet).

[Kai Pastor (dg0yt)]

The "existing structure" might be a reflection of the historical OGR being separate from the historical GDAL. Proliferating this now only to (perhaps) merge it later does not sound convincing.
And more features mean more testing mean more CI time.

[SunBlack (SunBlack)]

I just took a quick moment to look at their code. The old options seem to be broken anyway (mapnik/mapnik#4595), as far as I can tell, so we'll just use the combined one.

[Kai Pastor (dg0yt)]

Oh, another ogr?

[Billy O'Neal (BillyONeal)]

mapbox-polylabel removed the C++ code in mapbox/polylabel#124

Note that this isn't like "we removed the nice C++ wrappers around some C thing", this is a library where they had a C++ port of a Java thing in the same repo and they removed the C++ port. So there is no reason for C++ customers to expect support from the "original" implementation going forward, so we shouldn't fear the usual forms of conflict that come from forks.

(EDIT: To clarify, this is me confirming that I believe swapping in the fork is the correct thing to do)

[Billy O'Neal (BillyONeal)]

Drafting due to build failures, SunBlack#95 may fix.

Additionally, GPT 6 Sol reports:

• input-gdal, input-ogr, input-postgis, and input-pgraster are still advertised in vcpkg.json and mapped to old options in portfile.cmake, but upstream's combined-plugin PR changed Mapnik 4.3.2's dispatcher to add only the combined GDAL+OGR and PostGIS+PGRaster plugins (upstream source). Selecting any of these four features installs its dependencies but no corresponding plugin, although the feature tests pass.
• tiles.patch and mapnik-index.patch change general upstream build behavior, but no upstream issue or pull request is cited and repository searches found no submission. Submit the fixes upstream and cite them, or establish why they are vcpkg-specific, as required by the patching guidance.

[Billy O'Neal (BillyONeal)]

(I didn't ask it to try to fix the features report because you and Kai Pastor (@dg0yt) were discussing above and as not a user of this myself it's not clear what the correct change is)

[Billy O'Neal (BillyONeal)]

Looks like build is fixed but I'm still not sure what the right feature fix is.

[Copilot]

🟡 Changes recommended

Enabled datasource plugins are unavailable in static builds, MinGW utility linking is broken by an invalid libdl dependency, and the tiles feature omits a required redistribution license notice.

4 open issues

• Use CMAKE_DL_LIBS for MinGW compatibility · New
• Fix static GDAL/OGR and PostGIS/PGRASTER datasource registration · New
• Register the tiles datasource for static linkage · New
• Include mapnik-vector-tile license for input-tiles · New

What changed in this PR

Replaces mapbox-polylabel with Mapnik’s maintained C++ fork and updates Mapnik to 4.3.2 to use it.

Changes:

• Adds mapnik-polylabel 2.1.0 and removes the old port.
• Updates Mapnik dependencies and plugin features, including tiles support with optional SSL.
• Adds build patches and refreshes version records and viewer CI expectations.

Validation used the supplied diff and targeted upstream source inspection; no builds or consumer examples were run.

File	Description
versions/​m-/​mapnik.json	Records Mapnik 4.3.2.
versions/​m-/​mapnik-polylabel.json	Adds the fork’s initial version record.
versions/​baseline.json	Updates Mapnik and replaces the polylabel baseline.
scripts/​ci.feature.baseline.txt	Updates expected viewer feature failures.
ports/​mapnik/​vcpkg.json	Updates dependencies and plugin features.
ports/​mapnik/​usage	Normalizes whitespace without changing guidance.
ports/​mapnik/​tiles.patch	Makes tiles compile definitions target-scoped.
ports/​mapnik/​portfile.cmake	Updates source, feature mappings, downloads, and packaging.
ports/​mapnik/​mapnik-index.patch	Adjusts utility linkage on non-MSVC, non-Apple builds.
ports/​mapnik-polylabel/​vcpkg.json	Declares the renamed fork and version.
ports/​mapnik-polylabel/​portfile.cmake	Packages the fork’s headers and license.
ports/​mapbox-polylabel/​portfile.cmake	Removes the old Mapbox build recipe.

_{Review effort: Balanced}

[SunBlack (SunBlack)]

no idea what the AI want's here as the targets were not renamed, just the CMake option