On 30/07/2022 15:02, Ihor Radchenko wrote:
> 
> Hugo Heagren writes:
> 
>>> The rx-to-string call may still suffer from the described edge case.
>>
>> Yes, I think this is a good idea.
>>
>>> (and all-prefixes ...)
>>
>> I'm not sure what you mean, but I don't think it will work.
> 
> I mean
> 
> (and ,all-prefixes (rx-to-string `(: string-start (submatch (or (and ,@all-prefixes))) ":")))

I agree that it is better than the trick with test, but, I think, it is 
not critical taking into account version of the patch set. I do not see 
any issues that really prevent committing of changes. This time tests 
pass on my side.

When committing the current version or if you decided to prepare a new 
one, please, fix the name of the test. Ihor committed my patch to run ol 
tests, the variant that renames test instead of adjustment of selector.

diff --git a/testing/lisp/test-ol.el b/testing/lisp/test-ol.el
index 5f2cca538..6802f0333 100644
--- a/testing/lisp/test-ol.el
+++ b/testing/lisp/test-ol.el
@@ -673,7 +673,7 @@ API in `org-link-parameters'.  Used in test
  `:insert-description' is a function symbol."
    "foobar-from-function")

-(ert-deftest test-ol/insert-link-insert-description ()
+(ert-deftest test-org-link/insert-link-insert-description ()
    "Test `:insert-description' parameter handling."
    ;; String case.
    (should

To get impression of the new feature in action, I modified ol-info.el. 
First patch does not depend on :insert-description and might be useful 
per se. Second one just enables the feature.