bug#60841: 30.0.50; kill-ring-save pauses despite region being highlighted

[Eli Zaretskii]

> From: Kévin Le Gouguec <kevin.legouguec@gmail.com>
> Cc: gregory@heytings.org,  60841@debbugs.gnu.org
> Date: Mon, 23 Jan 2023 23:29:00 +0100
> 
> > seq.el is indeed preloaded, so that ship has sailed.  But you still
> > need to make sure seq is loaded _before_ any preloaded file which uses
> > it, and in this case faces is loaded before seq, so you cannot use
> > seq-difference.
> 
> (Thanks for spelling this out.  Do we have any documentation that calls
> out the precautions one must take when writing Elisp that will be
> preloaded, or any tooling that can detect whether some of those
> precautions were forgotten?  FWIW I saw no compiler warnings nor runtime
> errors with that patch)

Did you "make bootstrap"?  If not, some errors might not happen,
because the build will use previously compiled foo.elc files.

As for documentation: there's any number of such factoids related to
do's and dont's of Emacs development, and we lack a full-time
documentation fellow to keep all of them documented and up to date...

> >> +(defun region-highlighted-p ()
> >> +  "Say whether the region is visibly highlighted.
> >
> > Please drop the "Say" part, it's not our style.
> 
> ACK.  I see a few matches for "Return whether…" in-tree; would…
> 
>   Return whether the region stands out visually.
> 
> … be OK, or should I just go for…

It's OK, but IMO the "Return" part is almost redundant here.  But I
won't object to having it.

> "(elisp) Documentation Tips" recommends "Return t if", but merely as a
> way to "avoid starting the sentence with “t”", not because we have a
> preference for literally starting with "Return t if")

The point here is that this is a predicate, so it is known up front
that it will "return t" or nil.  The only non-trivial part is the
condition under which it will return non-nil.

Thanks.