netcdf: update to 4.9.3 (retry)

[remkos]

Description

Update to upstream version 4.9.3. This is a retry of PR #27638. See applicable comments there.

With this release, all the patches for this port have become obsolete.

Since the library version number changed, all the dependencies have been revbumped, except those that failed (xmgr, aircraft_oap, h4h5tools, metview, vapor). I leave it to others to fix those, since the failures are internal to the packages, not due to the netcdf update.

ncarg needed some extra patch files to get going, partially because of a routine being made obsolete.

The obsolete platforms darwin has been removed from the dependent Portfiles as well.

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 15.3.1 24D70 x86_64
Command Line Tools 16.2.0.0.1.1733547573

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL in commit message?
• checked your Portfile with port lint?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?
• checked that the Portfile's most important variants haven't been broken?

[macportsbot]

Notifying maintainers:
@nilason for port gdal.
@MarcusCalhoun-Lopez for port petsc, gerris.
@tenomoto for port cdo, ecCodes, esmf, gmt4, grads, magicspp, nco, netcdf-cxx, netcdf-cxx4, netcdf-fortran, netcdf, vis5d, wgrib2.
@barracuda156 for port elmerfem, mfem.
@thiagoveloso for port ncview.
@jwhowarth for port pymol.
@Veence for port gdal.
@Schamschula for port octave-netcdf.
@jswhit for port py-netcdf4.
@Dave-Allured for port nccmp, netcdf-fortran.
@dstrubbe for port paraview.
@jjstickel for port paraview.

[barracuda156]

Why R ports were failing? If that was simply a timeout (too much to build), that is not a stopper, but they still have to be revbumped.

[Dave-Allured]

-# Change github.tarball_from to 'releases' or 'archive' next update
-github.tarball_from         tarball
+github.tarball_from         archive

-checksums           rmd160  23fe2abab3f0d8249a93cfb08edaca31b32ff875 \
-                    sha256  c87e603447bb16a22df1a8df451885c32c32bcd5a6b0f73454ea9d1e2dc25c57 \
-                    size    25614625
+checksums           rmd160  0d5048ac645b4373372d538b8b3c92588da8ea3a \
+                    sha256  990f46d49525d6ab5dc4249f8684c6deeaf54de6fec63a187e9fb382cc0ffdff \
+                    size    25619216

[Dave-Allured]

netcdf portfile:
Remove obsolete user note. This is over 12 years old.

-notes "
-As of version 4.2 c++ and fortran interfaces are separate ports,\
-netcdf-cxx and netcdf-fortran, respectively.
-"
-

[remkos]

netcdf portfile: Remove obsolete user note. This is over 12 years old.

-notes "
-As of version 4.2 c++ and fortran interfaces are separate ports,\
-netcdf-cxx and netcdf-fortran, respectively.
-"
-

@Dave-Allured : Thanks for the revisions of netcdf/Portfile. I incorporated all of them.

[remkos]

Why R ports were failing? If that was simply a timeout (too much to build), that is not a stopper, but they still have to be revbumped.

In the end, only R-RNetCDF needed a minor tweek, then the others came along as well. So the three R packages (R-RNetCDF, R-easyNCDF and R-ncdf4) are now included!

[barracuda156]

@remkos Ah, that is another instance of a breakage caused by recent github.tarball_from tweaks, which were made without considering their effects.

Thank you!

[remkos]

With quite a bit of effort, I have also got ncarg and py-nio going.
The remaining dependent packages (xmgr, aircraft_oap, h4h5tools, metview, vapor), I did not manage to get going. The errors on which compilation failed were deep in the internals of the packages, usually in the code. Somebody else needs to resolve those.
I am not sure why the CI keeps failing, I managed to have all packages install on my machine.

[barracuda156]

Perhaps py-nio should increment revision to 29, not to 30.

[remkos]

Sorry, was a bit too radical increasing the revision number. Fixed now.

[Dave-Allured]

Three R ports are failing repeatedly on the 14_arm86 runner.

logs-macos-14.zip:

R-RNetCDF/logs/install-dependencies.log:Error: g95 is known to fail
R-easyNCDF/logs/install-dependencies.log:Error: g95 is known to fail
R-ncdf4/logs/install-dependencies.log:Error: g95 is known to fail

Ticket: R build fails due to g95 dependency from 4 months ago.

A quick read suggests that this might be different Xcode versions between remkos's system, and the 14_arm64 runner. Therefore I suggest that you drop these three ports from the current update, maybe try to bump them later in a separate PR.

[barracuda156]

It should be a CI-only failure (because of a broken Xcode). Just ignore it. Not revbumping ports will unnecessarily leave them broken.

[Dave-Allured]

It should be a CI-only failure (because of a broken Xcode). Just ignore it. Not revbumping ports will unnecessarily leave them broken.

That works for me. Thanks.

[remkos]

Thanks, @barracuda156 and @Dave-Allured, for your constructive comments.

I think from this we can conclude that we leave the three R packages in. Indeed, on Intel and MacOS 15 I get them nicely compiled.

For the other couple of ports that I did not manage to compile, I fear this is likely because the code is too outdated and someone needs to patch the code. For those ports I will assume to take the approach suggested by @Dave-Allured and put them in a later PR even though its CI is known to fail. But it will avoid confusing the outcome of the CI of the current PR even more.

I will prepare those revbumps and will submit them in a future PR as soon as this PR is merged.

[Dave-Allured]

@remkos The g95 problem for R ports is allegedly fixed. I suggest retrigger CI for this PR, see if it will pass now for macos-14.

https://trac.macports.org/ticket/71068#comment:28

[remkos]

@remkos The g95 problem for R ports is allegedly fixed. I suggest retrigger CI for this PR, see if it will pass now for macos-14.

https://trac.macports.org/ticket/71068#comment:28

Thanks, @Dave-Allured for that fix. I've kicked CI off again. Let's see.