Handle file-level docstrings in prog-mode

[doolio]

Closes #77.

This seems to work in a python file with a file level docstring.

As you'll see I really just adapted the existing beginend--point-is-in-comment-p (private) function. Not sure if I fully understand the let form yet so it may not be necessary in beginend--point-is-in-string-p. Plus you may not like this solution as it probably goes against the principles of the DRY methodology.

Not entirely sure how I should run the tests. I executed buttercup-run-at-point at the top level describe form in beginend-prog-test.el which resulted in the following output. Perhaps buttercup should have a buttercup-run-on-buffer type command as well. The individual test times are interesting. I added the new "for a docstring" test.

Running 9 specs.

beginend
  in prog-mode
    uses prog-mode-code-position-p to know where code begins (0.57ms)
    uses prog-mode-code-position-p to know where code end (0.39ms)
    moves point to beginning of first code line (0.77ms)
    moves point to end of last code line (0.57ms)
  beginend--prog-mode-code-position-p
    returns non-nil for a line of code (3.75ms)
    returns nil
      for a comment line (2.74ms)
      for a docstring (2.64ms)
      for an empty line (3.10ms)
      for a line with ^L (2.81ms)

Ran 9 specs, 0 failed, in 18.51ms.

The buttercup docs only document running tests when using cask, eldev, and eask but I think you use makel.mk right?

Since you use buttercup why do you need the .ert-runner file? Is it because you refer to ${TEST_ERT_FILES} in your Makefile. Excuse my ignorance here.

Finally, I noted the following comment at several places in begined-prog-test.el:

;; workaround for https://github.com/jorgenschaefer/emacs-buttercup/issues/84

But if understand the exchange in that issue then these comments are obsolete as you have put spy-on in a before-each form. I have left them in for now but can remove them if you agree.

[DamienCassou]

Thank you very much for your great work, I really appreciate.

Not sure if I fully understand the let form yet so it may not be necessary in beginend--point-is-in-string-p. Plus you may not like this solution as it probably goes against the principles of the DRY methodology.

Indeed, there is a lot of duplication but I think this is good enough.

Not entirely sure how I should run the tests.

Usually, you can check how the CI is ran in a project to see how to run the tests. In this case, the CI configuration is in .github/workflows. This will lead you to https://github.com/DamienCassou/emacs-workflows/blob/main/.github/workflows/makel.yml which last line is make check. Anyway, the CI checked your PR already and everything is fine!

I think you use makel.mk right?

Indeed: https://github.com/damienCassou/makel

why do you need the .ert-runner file?

This could be removed, you are right.

I have left them in for now but can remove them if you agree.

This comment should have been deleted some time ago. Feel free to get rid of them.

[DamienCassou]

I merged your PR. Feel free to open a new one for the cleanup you found.

[doolio]

I think you use makel.mk right?
Indeed: https://github.com/damienCassou/makel

It may not matter but I note your Makefile here is using an older version than your version on github.

[DamienCassou]

It may not matter but I note your Makefile here is using an older version than your version on github.

would you mind sending a PR to update makel?