Arch Linux User Repository

I finally got it installed again:

rm -rf /tmp/portmod.test/
sudo downgrade rust
> 1.79
yay -S python-redbaron
yay -S python-baron
yay -S portmod

Got me where I wanted. Thank you @bmwinger for your help.

I have the following insights:

• For some reason redbaron and the baron package were not properly resolved as dependency for me
• There was no hint for me visible that the conflicting path was /tmp/portmod.test/, hence it would be nice to be warned about this.

<3

That's something that can be prevented in portmod, as it should be cleaning up that directory automatically (currently it won't if an error occurs in a certain part of the code), but you can also do it yourself by deleting /tmp/portmod.test, where it's storing the temporary prefix used for generating the CLI docs.

Thank you for your feedback.

I am still unable to install it.

sudo downgrade rust > 1.79

yay -S portmod

...
writing output... 
building [man]: all manpages
updating environment: [new config] 47 added, 0 changed, 0 removed
reading sources... [  9%] cli/portmod
Exception occurred:
  File "/home/XXX/.cache/yay/portmod/src/portmod-v2.7.3/portmod/prefix.py", line 125, in add_prefix
    raise PrefixExistsError(prefix)
portmod.prefix.PrefixExistsError: The prefix <example> already exists
The full traceback has been saved in /tmp/sphinx-err-601d54bt.log, if you want to report the issue to the developers.
Please also report this if it was a user error, so that a better error message can be provided next time.
A bug report can be filed in the tracker at <https://github.com/sphinx-doc/sphinx/issues>. Thanks!
make: *** [Makefile:20: man] Error 2
make: Leaving directory '/home/XXX/.cache/yay/portmod/src/portmod-v2.7.3/doc'
==> ERROR: A failure occurred in build().
    Aborting...
 -> error making: portmod-exit status 4
 -> Failed to install the following packages. Manual intervention is required:
portmod - exit status 4

@Dornhoeschen: I see. It looks like python-autodocsumm is orphaned and portmod is the only thing depending on it, so I should probably adopt it and give it an update. I guess the git tag changed on the upstream source or something, because the package seems like it should still work.

The issue with building time is resolved with newer versions, and doesn't appear to be an issue with portmod 2.8. However unfortunately that fails later due to not linking zstd-rs properly (required by tantivy, which is used to index packages). The portmod 2.8 build works fine everywhere else though, it's just failing with the arch build system.

Unfortunately memory safety has nothing to do with reliable builds. Other compilers break things on updates too (e.g. the gcc 14 update caused build failures for me in other software). The issue here is really just that we're doing an old build with a new compiler. It should work fine on <1.80, so you can work around the issue by installing rustup and rustc 1.79, and using that to build portmod 2.7.3.

Actually, only thing depending on the time crate is also tantivy, which is portmod's largest rust dependency, and is a big and complex project. Given that it's not particularly integral and has been causing problems I'm tempted to drop it for the sake of maintainability. But that's more of a portmod discussion than an AUR discussion.

Currently broken on my end.

The dependency integrity error on python-autodocsumm can be resolved with yay -S --mflags --skipinteg portmod.

The above error might also be caused by the fact that an outdated version of python-autodocsumm is in AUR due to the lack of a maintainer.

But then I run into error: could not compiletime(lib) due to 1 previous error which seems to be an "small" breaking change in rust's API according to this: https://discourse.nixos.org/t/rustc-explain-e0282-error-could-not-compile-time-lib/50765

If you allow this sarcastic comment: The packages which break the most are those with the super secure rust language - caused by "small" breaking changes. From rusts website: A language empowering everyone to build >reliable< and efficient software. Its obviously false advertisement

If you, the kind maintainer, could look into that, I'd appreciate it <3

@bmwinger Thanks! I actually didn't know that. I was able to skip the checks just fine and upgraded. For anybody else with the same issues, do a reinstall of portmod and all dependencies, commenting out the check sections in the PKGBUILDs of the python packages that fail.

@RamenField: See https://wiki.archlinux.org/title/Python#Module_not_found_after_Python_version_update. This is a standard issue with how the AUR interacts with arch's binary packages which everyone using the AUR needs to be aware of.

The only thing that failed for me was redbaron, and that was just a test for the optional pygments functionality, so I just skipped the tests for redbaron when installing it.

Dependency python-rply (orphaned) will fail to build since last portmod update. Installing python-rply-git builds, but causes a build failure in portmod:

Configuration error:
There is a programmable error in your configuration file:

Traceback (most recent call last):
  File "/usr/lib/python3.11/site-packages/sphinx/config.py", line 354, in eval_config_file
    exec(code, namespace)  # NoQA: S102
    ^^^^^^^^^^^^^^^^^^^^^
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/doc/conf.py", line 19, in <module>
    from portmod.globals import get_authors, get_version  # noqa
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmod/__init__.py", line 4, in <module>
    import portmod.globals  # noqa # pylint: disable=unused-import
    ^^^^^^^^^^^^^^^^^^^^^^
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmod/globals.py", line 150, in <module>
    from portmod.repos import get_local_repos  # noqa: E402
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmod/repos.py", line 14, in <module>
    from .config import get_config
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmod/config/__init__.py", line 25, in <module>
    from .pyconf import (
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmod/config/pyconf.py", line 14, in <module>
    from portmodlib._loader import MINIMAL_GLOBALS, Policy
  File "/tmp/makepkg/portmod/src/portmod-v2.6.2/portmodlib/_loader.py", line 19, in <module>
    from RestrictedPython import (
ModuleNotFoundError: No module named 'RestrictedPython'

make: *** [Makefile:20: man] Error 2
make: Leaving directory '/tmp/makepkg/portmod/src/portmod-v2.6.2/doc'
==> ERROR: A failure occurred in build().
    Aborting...
 -> error making: portmod-exit status 4

Required dependencies got abandoned and they don't build :(

error: can't build portmod-2.6.2-1, deps not satisfied: python-redbaron
error: packages failed to build: python-rply-0.7.8-4  python-baron-0.10.1-2  python-redbaron-0.9.2-8  portmod-2.6.2-1

orphaned deps: python-redbaron, python-baron, python-rply

Arf the issue was submodule.recurse = true in my user git config. The tests includes running the git CLI to fetch a repo, and it loads my config.

As a result, running git pull -v --rebase -- origin somehow fails. Indeed, for some reasons --recurse-submodules and -v appear incompatible with my git version (2.39.2).

Looks like this: https://lore.kernel.org/git/xmqqiliur6t9.fsf@gitster.g/T/