Slawomir Grochowski <[email protected]> writes:
> You are right. I should have marked the test explicitly. We discussed
> this earlier, but did not reach a final decision on the exact
> convention, so I hesitated.
>
> I have now added the characterization prefix, as proposed in
> <https://list.orgmode.org/[email protected]/> , and the following
> disclaimer:
>
> NOTE: This pins a known arbitrary behavior to prevent regressions.
> It does NOT represent the intended design spec.
If you're going to mark that test, you should also mark the test that
sparked the original discussion as well:
`test-org-colview/insert-columns-keyword'
I also want to ask a question about test naming though. What is the
standard for naming tests? This is actually a pretty important question
I think because that's used for ERT selectors.
It seems to be something like this with many exceptions:
#+begin_src elisp
(rx
(optional "test-")
(literal FILE-NAME)
(or "-" "/")
(literal (string-trim-left FUNCTION-NAME
(rx (literal FILE-NAME) (one-or-more "-"))))
(optional
(or "-" "/")
(literal EXTRA-DESCRIPTOR)))
#+end_src
Personally, I think the current formula is a little lax.
Should the beginning "test-" be optional or do we want to fix that? Is
it redundant information that can simply be removed?
Are we ok with the file-name separator being "[-/]" or should we
standardize on "/"?
I don't think we should worry too hard about trying to standardize the
second half of the test name.
Also where do we want to shove in the word 'characterization'? Would it
be better not not shove it in and just make it a tag?
Should we change the default ERT selector to exclude characterization
tests? I believe we should to prevent new contributors from fighting
frivolous noise.
