Howard Chu wrote:

I guess we need a syntax flag that says "denormalization needed" and perhaps a usage flag for the normalizer that says "denormalize a normalized value"

Rather than "syntax", a "matching rule" flag, as normalization is per matching rule. UUID matching rules already have a SLAP_MR_MUTATION_NORMALIZER that prevents UUIDs from being normalized when used in a DN. Probably we need to do something slightly different, as the current approach would not normalize BADBADEF-DBAD-1029-92f7-BADBADBADBAD into badbadef-dbad-1029-92f7-badbadbadbad unless dnPrettyNormal is called: in that case, pretty would lowercase the A-F chars, so normal would see them lowercased. For example:

$ slapdn 'entryUUID=BADBADEF-DBAD-1029-92f7-BADBADBADBAD' -N entryUUID=BADBADEF-DBAD-1029-92f7-BADBADBADBAD $ slapdn 'entryUUID=BADBADEF-DBAD-1029-92f7-BADBADBADBAD' -P entryUUID=badbadef-dbad-1029-92f7-badbadbadbad $ slapdn 'entryUUID=BADBADEF-DBAD-1029-92f7-BADBADBADBAD' DN: <entryUUID=BADBADEF-DBAD-1029-92f7-BADBADBADBAD> check succeeded normalized: <entryUUID=badbadef-dbad-1029-92f7-badbadbadbad> pretty: <entryUUID=badbadef-dbad-1029-92f7-badbadbadbad>

which sounds a bit inconsistent (no issue in slapd: DNs coming from client requests are always pretty-normalized, while DNs for internal use are always in pretty or normal form).

So I'd go for using the SLAP_MR_MUTATION_NORMALIZER flag to determine if a value needs to be de-normalized, and OR it to the usage in normalization hooks (smr_normalize) to instruct them to de-normalize rather than normalize values.

p.

Ing. Pierangelo Masarati OpenLDAP Core Team

SysNet s.r.l. via Dossi, 8 - 27100 Pavia - ITALIA http://www.sys-net.it --------------------------------------- Office: +39 02 23998309 Mobile: +39 333 4963172 Email: pierangelo.masarati@sys-net.it ---------------------------------------