None of my questions are answered.
And I do not understand the output of your Microsoft GitHub CI system in the back.

The repo does miss contributor documentation for things like this.
I could fix it all my self if I could understand it.

Where in the HTML output is that hardlink HTML file used?

It is not directly used by pydoctor: it’s designed like this to keep compatibility with old links that refers to the root module name. So all modules can be accessed by their full name.

Regarding the contributors documentation.. please keep in mind that we’re all volunteers here. Best documentation will always be to read the code.

For the branch model: ne cherches pas midi à 14:00 like we say French: just target the master branch for your PRs.

Regarding the contributors documentation.. please keep in mind that we’re all volunteers here. Best documentation will always be to read the code.

I am a dev and a maintainer, too. But beeing a volunteer is IMHO not good excuse for not documenting. Code is not documentation. And contributors documentation does not have something to do with code by the way. It is more about how to project is organized and how to do PRs, how the test/build/CI system is intended to work etc pp.

But lets stop the meta discussion and lets go back to this PR.
How am I suppose to understand the failing Microsoft GitHub Action things?

How can I help you to fix this PR and get it merged.

How am I suppose to understand the failing Microsoft GitHub Action things?

First, your changes makes pydoctor crash. The cause of this must still be investigated.

Secondly, there is a bug in pydoctor_primer that crashes. I’m not sure whether it’s because the CI step is only working for privileged users or because the diff to show is too big because the change is making pydoctor crash and the diff is then too large.

How can I help you to fix this PR and get it merged.

We need to examine the stack trace and fix the bug. It might work on your environment because your already have a html build existing so your not seeing the crash. Please try to delete the output directory and see if pydoctor still works.

First, your changes makes pydoctor crash. The cause of this must still be investigated.

I don't see a crash. Can you link to the output in some of that Microsoft GitHub Actions CI jobs/machines?
What really annoys me about the CI is that there are multiple failing "jobs" (?). The CI should stop at the first error and not overwhelming contributors. I have not enough information (documentation!) to decide which one of that CI jobs is the most relevant I should focus my resources on.

Secondly, there is a bug in pydoctor_primer that crashes.

I do not know what that is and it is not documented somewhere and not even in the primar yaml itself.
You would save yourself and future contributors a lot of time and words if you would document it.

We need to examine the stack trace and fix the bug.

already have a html build existing so your not seeing the crash. Please try to delete the output directory and see if pydoctor still works.

In that case I can reproduce this problem.

/home/user/ownCloud/my.work/buhtzology/docs
Traceback (most recent call last):
  File "/home/user/ownCloud/my.work/buhtzology/docs/./hey_doc.py", line 152, in <module>
    rc = pydoctor.driver.main(args)
  File "/home/user/ownCloud/my.work/pydoctor/pydoctor/driver.py", line 169, in main
    make(system)
  File "/home/user/ownCloud/my.work/pydoctor/pydoctor/driver.py", line 128, in make
    writer.writeSummaryPages(system)
  File "/home/user/ownCloud/my.work/pydoctor/pydoctor/templatewriter/writer.py", line 116, in writeSummaryPages
    os.link(root_module_path.parent / 'index.html', root_module_path)
FileNotFoundError: [Errno 2] No such file or directory: 'html/index.html' -> 'html/buhtzology.html'

Again: I can not see this error in your CI or tests. How can an extern but motivated contributor can see this? You should document this. Your project is not ready for extern contributors because PRs and even testing is possible only with project-intern knowledge.

You can blame me by the way: Your test suite do catch the problem very well. I just haven't run the tests on my local machine.

=================================================== short test summary info ===================================================
FAILED pydoctor/test/test_commandline.py::test_main_symlinked_paths - FileNotFoundError: [Errno 2] No such file or directory: 'apidocs/index.html' -> 'apidocs/basic.html'
FAILED pydoctor/test/test_commandline.py::test_main_source_outside_basedir - FileNotFoundError: [Errno 2] No such file or directory: 'apidocs/index.html' -> 'apidocs/basic.html'
ERROR pydoctor/test/epydoc/test_pyval_repr.py::test_expressions_parens

About the tests

I remember why I didn't run the tests. You use the real filesystem instead of a virtual filesystem for your tests (see #721).

About pydoctor/test/epydoc/test_pyval_repr.py::test_expressions_parens I opened #754.

About pydoctor/test/test_commandline.py::test_main_symlinked_paths and ::test_main_source_outside_basedir. I am still investigating that two failing tests...
I would say these are system tests. My recommendation is to reorganize the test files and separate them between "unit", "integration" and "system" tests. It improves understanding for external contributors. Additionally I would suggest to locate the "test" folder outside of the source-package folder. (e.g. using the "src Layout").

Can you link to the output in some of that Microsoft GitHub Actions CI jobs/machines?

https://github.com/twisted/pydoctor/actions/runs/7346693679/job/20001879420?pr=722

The CI should stop at the first error and not overwhelming contributors. I have not enough information (documentation!) to decide which one of that CI jobs is the most relevant I should focus my resources on.

The most important is the unit tests, you can disregard the pydoctor_primer failure at this time.

You should document this. Your project is not ready for extern contributors because PRs and even testing is possible only with project-intern knowledge.

I started contributing to pydoctor in 2020 when they had no documentation whatsoever. Our contributor documentation talk about using tox for testing, it doesn’t talk about the primer because it’s something that I just set up. You can look at this PR to understand the goal of the primer, that is to comment PRs and show the diff in the pydoctor log: #739 (comment)

I will investigate this problem when I have some time next week, in the mean time I’ll try to answer your questions

The solution is not that simple. I tried

def make(system: model.System) -> None:
    """
    Produce the html/intersphinx output, as configured in the system's options. 
    """

    # ...

        if options.htmlsubjects:
            subjects = [system.allobjects[fn] for fn in options.htmlsubjects]
            writer.writeIndividualFiles(subjects)
        else:
            writer.writeIndividualFiles(subjects)
            writer.writeSummaryPages(system)

Other tests then fail. (btw: I miss the opportunity to run a single test only. Not sure how to do this with tox. Pytest alone still do not work.)

I am not enough into the code to understand it and to decide about a solution. My suggest would be refactor writer.py::TemplateWriter.writeSummaryPages() and driver.py::make(). Especially the first one is to complex and IMHO could be broken down into multiple simpler methods.

In the current state my understanding of the code and the logic behind it is not deep enough to find a solution myself.

The test summary

FAILED pydoctor/test/test_commandline.py::test_main_return_non_zero_on_warnings - assert 0 == 3
FAILED pydoctor/test/test_templatewriter.py::test_basic_package - FileNotFoundError: [Errno 2] No such file or directory: '/tmp/pytest-of-user/pytest-3/test_basic_package0/basic.html'

It seems that the original pydoctor/test/test_commandline.py::test_main_symlinked_paths and ::test_main_source_outside_basedir passed.