I would like to read user_prefs from a LDAP server. How to setup the
connection etc. has been documented, but I miss the bit that tells how to
store the configuration.

If my LDAP knowledge doesn't deceive me, then the spamassassin attribute shown
in the README is multi-valued, but this ability is not used.

Instead - as far as I understand it - the whole configuration goes as one big
chunk into the spamassassin attribute.

Following my understanding an example, that adds headers to the mail in any
case and whitelists sender@example.com as well as sender@example.net,
blacklists nasty@example.com and gives a score of 10.0 to SUBJ_ILLEGAL_CHARS
looks like this:

dn: cn=Curley Anderson,ou=MemberGroupB,o=stooges
spamassassin: add_header all \
whitelist_from sender@example.com \
whitelist_from sender@example.net \
blacklist_from nasty@example.com \

Is my understanding correct? If not, how would it be done?