bug#64089: 30.0.50; `ldap-search' errors out with `wrong-type-argument listp' when called WITHDN == t

[Eli Zaretskii]

> Date: Sun, 18 Jun 2023 10:51:25 +0200
> From: Jens Schmidt <jschmidt4gnu@vodafonemail.de>
> Cc: 64089@debbugs.gnu.org, monnier@iro.umontreal.ca
> 
> On 2023-06-18  09:43, Jens Schmidt wrote:
> 
> > Or whatever.  In any case, I'll open a new bug for that to continue 
> > this discussion.
> 
> And of course, Filipp is right w.r.t. to API consistency.  TBH, I simply
> haven't payed attention to `ldap-ignore-attribute-codings' until
> recently.  So for emacs-29, I can provide a patch that backs out my
> previous patch and fixes the bug such that the *original/Gerd's* string
> 
>    "dn: cn=JENS_SCHMIDT,L=REGION,DC=COMPANY,DC=COM"
> 
> gets appended to the result w/o triggering the
> 
>    wrong-type-argument listp
> 
> for both `ldap-ignore-attribute-codings' equaling nil and non-nil.
> 
> Basicaly the upper half of Filipp's patch, but redone such that even I
> understand it without too much eye-balling :-).  Like this probably (not
> tested yet!):
> 
>      (cond
>       (ldap-ignore-attribute-codings
>        result)
>       (withdn
>        ;; Do not process first elements of the records
>        ;; with `ldap-decode-attribute' (Bug#64089).
>        (mapcar (lambda (record)
>                  (cons (car record)
>                     (mapcar #'ldap-decode-attribute
>                                (cdr record)))
>               result)))
>       (t
>        (mapcar (lambda (record)
>               (mapcar #'ldap-decode-attribute record))
>                  result))))
> 
> WDYT?

I'm uncomfortable with non-trivial changes, so if this is anywhere
like Filipp's proposal, I'd rather leave it unfixed on emacs-29.