Fix generate-conda-packages job

[traversaro]

Fix #1572 .

Specific fixes:

• Pin swig to 4.1.0 for conda packages generation

[traversaro]

Great, pinning swig 4.1.0 creates other problems:

RuntimeError: Solver could not find solution.Mamba failed to solve:
 - ipopt 3.14.13.*
 - swig 4.1.0.*
 - qt-main 5.15.*
 - libopencv 4.9.0.*
 - icub-firmware-shared
 - gsl 2.7.*
 - xorg-libxfixes
 - ycm-cmake-modules
 - sdl
 - python 3.8.* *_cpython
 - libyarp
 - libode
 - libglu
 - libstdcxx-ng >=12
 - ace
 - libdc1394
 - libgcc-ng >=12
 - freeglut

with channels:
 - local
 - conda-forge
 - robotology

The reported errors are:
- Encountered problems while solving:
-   - package swig-4.1.0-hb8327b8_3 requires pcre2 >=10.37,<10.38.0a0, but none of the providers can be installed
- 

[traversaro]

See https://github.com/robotology/robotology-superbuild/actions/runs/7666685819/job/20894997057 .

[traversaro]

Ok, we have a more clear situation with mamba install:

C:\Users\straversaro>set CONDA_SUBDIR=linux-64
C:\Users\straversaro>set CONDA_OVERRIDE_GLIBC=2.30

C:\Users\straversaro>mamba create -n testswig swig=4.1.0 ipopt==3.14.13 qt-main=5.15 libopencv==4.9.0 gsl==2.7.* sdl libyarp libode libglu ace libdc1394

                  __    __    __    __
                 /  \  /  \  /  \  /  \
                /    \/    \/    \/    \
███████████████/  /██/  /██/  /██/  /████████████████████████
              /  / \   / \   / \   / \  \____
             /  /   \_/   \_/   \_/   \    o \__,
            / _/                       \_____/  `
            |/
        ███╗   ███╗ █████╗ ███╗   ███╗██████╗  █████╗
        ████╗ ████║██╔══██╗████╗ ████║██╔══██╗██╔══██╗
        ██╔████╔██║███████║██╔████╔██║██████╔╝███████║
        ██║╚██╔╝██║██╔══██║██║╚██╔╝██║██╔══██╗██╔══██║
        ██║ ╚═╝ ██║██║  ██║██║ ╚═╝ ██║██████╔╝██║  ██║
        ╚═╝     ╚═╝╚═╝  ╚═╝╚═╝     ╚═╝╚═════╝ ╚═╝  ╚═╝

        mamba (1.4.2) supported by @QuantStack

        GitHub:  https://github.com/mamba-org/mamba
        Twitter: https://twitter.com/QuantStack

█████████████████████████████████████████████████████████████


Looking for: ['swig=4.1.0', 'ipopt==3.14.13', 'qt-main=5.15', 'libopencv==4.9.0', 'gsl=2.7', 'sdl', 'libyarp', 'libode', 'libglu', 'ace', 'libdc1394']

conda-forge/linux-64                                        Using cache
warning  libmamba Cache file "C:\\Users\\straversaro\\AppData\\Local\\miniforge3\\pkgs\\cache\\09cdf8bf.json" was modifmconda-forge/noarch                                  15.4MB @   5.9MB/s  3.1s
Could not solve for environment specs
The following packages are incompatible
├─ libopencv 4.9.0  is installable and it requires
│  └─ libglib >=2.78.3,<3.0a0 , which requires
│     └─ pcre2 >=10.42,<10.43.0a0 , which can be installed;
└─ swig 4.1.0**  is uninstallable because there are no viable options
   ├─ swig 4.1.0 would require
   │  └─ pcre2 >=10.40,<10.41.0a0 , which conflicts with any installable versions previously reported;
   └─ swig 4.1.0 would require
      └─ pcre2 >=10.37,<10.38.0a0 , which conflicts with any installable versions previously reported.

So we have an incompatibility between libopencv and the swig we want to pin. Interestingly, this does not happen in YARP in conda-forge, probably because there swig is a build dependency, and so the build and host dependencies use different environments. Probably we can also keep swig as build dependency also in the robotology-superbuild, to fix the problem.

[traversaro]

Probably we can also keep swig as build dependency also in the robotology-superbuild, to fix the problem.

Done in c392de8 .

New test build: https://github.com/robotology/robotology-superbuild/actions/runs/7667984356 .

[traversaro]

The windows build was successful: https://github.com/robotology/robotology-superbuild/actions/runs/7666685819/job/20894997469 . So I guess also the Linux and macOS build will be successful, if the swig fix (that was not necessary on Windows) actually works.

[traversaro]

So I guess also the Linux and macOS build will be successful, if the swig fix (that was not necessary on Windows) actually works.

The swig fix worked. Let's merge, so we can actually generate the packages. We can fix further failures in future PRs.