Regarding:  no91-3568 (Elyakim  ben  Avraham, mi-London, 1745-1814):
This author wrote only  in Hebrew under  the  name chosen as  the heading.
This accords with rule 22.2B1:  "If  all  the works  by one  person appear  under  one pseudonym, choose  the pseudonym."
The rule goes on  to  say that if  the real name is known, a  ref  should be made from it,  and this has been done (there is  a  ref
from Hart, Jacob, 1745-1814).  There might  be  some argument here  as  to which is  the "real name" and which is  the
"pseudonym," but anyway the heading is  the  name under which the  books are written, which is the critical point.

------------------------------

Another case deals with Sh. Didovski who writes under the penname Dido. The NAF already has another Dido:

010     no 99-062911
100 0   Dido
400 1   Armstrong, Dido
400 1   Armstrong, D.  (Dido)
670     No angel, 1999:  label (Dido) booklet (D. Armstrong)
670     Time, June 7, 1999:  p. 78 (Dido, 26-year-old pop singer)
670     US copyright file, Aug. 25, 1999 (Armstrong, Dido; Dido)

1.  LCRI 22.2B (p. 1, point 3 under "Multiple Headings--Contemporaries") says that when a contemporary person publishes under at least one pseudonym, there must be a heading for thhe name published under (here, "Dido" ) and ALSO a heading for the "possibly unused" real name (here, "Didovski, Sh.").

2.  If there's already a heading identical to one of these headings, and they can't be distinguished by one of the legitimate kinds of qualifiers ($d, $q, or some kinds of $c), the existing heading must become "undifferentiated" or "non-unique," with the various changes in coding and special 670s that process entails.

3.  Legitimate $c-type qualifying terms for 20th-century persons are described in rule 22.19B1 ("e.g., terms of address, titles of position or office, initials of an academic degree, initials denoting membership in an organization"--but see also in paragraph 5 below), and "poet" is not such a term.  There are some cases in the database where "poet" has beeen used, but these are not justified by rules.  CPSO  specifically states that these are not legitimate.

4.  LCRI  22.17-22.20, p.2 allows for $q qualifiers which are not simple expansions of initialisms, with examples like "Clapper, John$q(John Samuel)" and "Jorgen, Dick$q(Richard Clark)."  But CPSO distinguishes cases like these from
"Dido$q(Sh. Didovski)" on the grounds that the two elements on either side of the $q in the proposed example are too dissimilar.

5.  CPSO points out, though, that rule 22.19A1 allows for a made-up parenthetical qualifier to break conflicts if the entry element is a given name.
What's NOT allowed is "Dido,$cpoet" (rule 22.19B1).
What IS allowed is "Dido$c(Poet)" (rule 22.19A1), but this is allowed only because (or if) "Dido" is a "forename."
We are not sure that "Dido" is a given name, but on the other hand, what else could it be?
So how about "Dido$c(Poet)" for the heading?  Then it won't have to be non-unique.