from NFC to NFKC in UTF8bvnormalize() and UTF8bvnormcmp()

Re: Untested patch: back-tcl used wrong types  (ITS#1719)

			================
Written by Hallvard B. Furuseth and placed into the public domain.
This software is not subject to any license of the University of Oslo.
			================

> I turned it into an automatic variable.

...and used a variable-length array.  That's a gcc extension, it is not
in ANSI C89.  (It is in C99 though.)  You seem to be compiling without
-pedantic:-)  Anyway, here is a patch to turn it back into ch_malloc(),
plus some README fixes

  deleted ldap_pvt_tls_get_peer()
  changed ldap_pvt_tls_get_peer_dn() to use ldap_X509dn2bv()
  added ldap_pvt_tls_get_my_dn()

normalized LDAP DNs now.

Changed dnDCEnormalize to dnX509normalize, added dnX509peerNormalize,
based on new ldap_X509dn2bv() etc.

working ifor Cyrus 2. (Both GSSAPI and EXTERNAL are broken at the moment.)

Here a patch for another:  A spurious `"' in back-ldbm.
From Hallvard.

Added LDAP_LOG Messages

Added LDAP_LOG Messages

after v3refs have been chased. They are fully processed by the time we get
back, so we should just return the current result message to the caller.

dn_openssl2ldap() and sasl_external_x509dn_convert. Fixed realm handling
for foreign Kerberos realms embedded in usernames.

by ITS#1360.

			================
Written by Hallvard B. Furuseth and placed into the public domain.
This software is not subject to any license of the University of Oslo.
			================

Bug fixes:
- acl_regex_normalized_dn(pattern):
  * used pattern->bv_len even though it claimed not to,
  * would walk past the end of strings that ended (incorrectly)
    with a single '\'.
- style=regex checked for "^.*$$" twice but not for "^.*$".
- the code did not notice if dnNormalize2() failed, and would
  (at least in one case) treat a bad DN as '*'.
Some cleanup:
- changed regtest() to return void, since the return value was unused.
- changed acl_regex_normalized_dn() to take a string input argument
  instead of a half-filled berval, it looks saner that way.

Hallvard B. Furuseth <h.b.furuseth@usit.uio.no>, April 2002.

			================
Written by Hallvard B. Furuseth and placed into the public domain.
This software is not subject to any license of the University of Oslo.
			================

Here are fixes for more places where the argument to ctype.h functions
should be in the range of `unsigned char'.

Explanation of the last patch (to schema_init.c:bvcasechr()):
TOLOWER() and TOUPPER() return values in the range of `unsigned char',
but bvcasechr() then compares those values with a plain `char'.  So I
convert the return values from TOLOWER()/TOUPPER() to `char' first.

Hallvard B. Furuseth <h.b.furuseth@usit.uio.no>, April 2002.

Added LDAP_LOG Messages

			================
Written by Hallvard B. Furuseth and placed into the public domain.
This software is not subject to any license of the University of Oslo.
			================

ucgendat.c accessed unallocated memory when i == ncodes_size.

The changes others are trivial, I just include them since I'm patching
ucdata anyway:

ucdata.c   had some pointless '0 <= unsigned' comparisons.

ucstr.c    assigned a long* to an unsigned long*.  Since malloc()
           returns void*, the result need not be cast at all.

I'll send the ucgendat.c and ucdata.c patches to Mark Leisher
<mleisher@crl.nmsu.edu>.

Hallvard B. Furuseth <h.b.furuseth@usit.uio.no>, April 2002.

Added LDAP_LOG Messages