bug#66649: 29.1; `project-remember-projects-under' behavior doesn't match its doc

[Philip Kaludercic]

Damien Cassou <damien@cassou.me> writes:

> Dmitry Gutov <dgutov@yandex.ru> writes:
>> On 02/11/2023 21:58, Damien Cassou wrote:
>>> If I may, the code of `project-remember-projects-under' seems to suffer
>>> from a similar mismatch between the docstring and implementation when
>>> RECURSIVE is nil: only the DIR directory is tested and not "known
>>> projects below a directory DIR".
>>
>> Have you seen this problem when testing it?
>
> The patch is working perfectly find, thank you. I was talking about an
> unrelated (but similar) problem in a different function:
> `project-forget-projects-under'. My message incorrectly referred to the
> function you already fixed. Sorry for the confusion.

I don't think this is the same problem, in
`project-forget-projects-under' there is no manual recursive descent,
just some duplicated code.  We could also re-write it to look like this:

diff --git a/lisp/progmodes/project.el b/lisp/progmodes/project.el
index 95db9d0ef4c..5f1cce160b2 100644
--- a/lisp/progmodes/project.el
+++ b/lisp/progmodes/project.el
@@ -1905,15 +1905,12 @@ project-forget-projects-under
 forgotten projects."
   (interactive "DDirectory: \nP")
   (let ((count 0))
-    (if recursive
-        (dolist (proj (project-known-project-roots))
-          (when (file-in-directory-p proj dir)
-            (project-forget-project proj)
-            (setq count (1+ count))))
-      (dolist (proj (project-known-project-roots))
-        (when (file-equal-p (file-name-directory proj) dir)
-          (project-forget-project proj)
-          (setq count (1+ count)))))
+    (dolist (proj (project-known-project-roots))
+      (when (if recursive
+                (file-in-directory-p proj dir)
+              (file-equal-p (file-name-directory proj) dir))
+        (project-forget-project proj)
+        (setq count (1+ count))))
     (if (zerop count)
         (message "No projects were forgotten")
       (project--write-project-list)

But that would incur a branch in every iteration of `dolist'.

Either way, I'll push the first patch to master since there haven't been
any objections to that change.