Author: sparkins@redhat.com
Summary: Re-implemented bug-fix to add a new constructor for the JSSESocketFactory class that allows an explicit socket factory to be passed in. This bug-fix was reimplemented in order to re-license the LDAP JDK under the standard Mozilla tri-license.
Author: sparkins@redhat.com
Summary: Re-implemented two bug fixes in the LDIFWriter class in order to allow us to relicense the LDAP JDK under the standard Mozilla tri-license. The first bug had to do with the printEntryStart() method of the LDIFWriter class does not handling non-printable DN's properly. The second bug had to do with only the first value of a multi-valued attribute being written out in LDIF format.
Author: sparkins@redhat.com
Summary: Re-implemented two bug fixes in the LDIFWriter class in order to allow us to relicense the LDAP JDK under the standard Mozilla tri-license. The first bug had to do with the printEntryStart() method of the LDIFWriter class does not handling non-printable DN's properly. The second bug had to do with only the first value of a multi-valued attribute being written out in LDIF format.
Author: sparkins@redhat.com
Summary: Re-implemented bug-fix to add a new constructor for the JSSESocketFactory class that allows an explicit socket factory to be passed in. This bug-fix was reimplemented in order to re-license the LDAP JDK under the standard Mozilla tri-license.
Description: nsldif32.tdf is the template file; nsldif32.def is a derived file in the autotool build. It needs to be in CVS not to break the old style build.
Description: Windows: libldap and the LDAP tools should support SASL/Digest
Fix Description: The biggest change was to hack sasl.m4 to make AC_CHECK_LIB work on Windows. I
had to override LIBS to be able to pass in /link /LIBPATH:c:/path/to/sasl
libsasl.lib, libsasl2.lib, and sasl32.lib.
Another big change was to implement getlogin() on Windows, required by
ldaptool-sasl.c. The implementation is pretty simple.
There were several places in the Makefiles that assumed Windows meant no
support for SASL, so I just changed them, and added some support for SASL_LIBS
in some places where it was missing.
Description: Make libldif public - windows specific fixes
Fix Description: Windows needed a nsldif32.tdf file in order to generate
the nsldif32.def file using the libldif.ex file (committed in the previous
commit).
Also, it has been a while since the .def files were generated. The libX.ex
files have had quite a few changes, and we needed to regenerate all of the
.def files.
Finally, this will be version 6.0.2 of the mozilla ldap c sdk.
Description: Make libldif public - fix Solaris build
Fix Description: libldif
needs to specify the exports. I added a libldif.ex and fixed
libldif/Makefile.in to correctly generate and use the exports listed in the
file.
Description: Make libldif public
Fix Description: Copy/paste the makefile code to build the shared library in addition to the static library from libldap/Makefile.in. The build will now build both libldif60.so and libldif60.a, and package/distribute both of them. I changed include/Makefile.in to export ldif.h with the other public header files, and changed ldif.h to add a blurb explaining that these interfaces were formerly private and use with caution.
Since this is a non trivial change, I bumped the version number to 6.0.1.
Description: Change LDAP C SDK version to 6
Fix Description: Change the version to major 6, minor 0, and submin/patch 0. The change is in various places in the build makefiles and header files. The shared lib suffix is now 60 (from 50).
In a somewhat unrelated change, the configure script had to change to allow
building with the RPM spec %configure macro on rhel/fedora. That macro
defines --host, --build, and --target, and defines --host different than
--target. This caused configure to think it was doing a cross compilation.
Instead, the correct cross compilation trigger is if --host is specified and
it is different than --build. However, the mozilla builds expect to be able
to pass in --host different than --target, so all of these different
combinations had to be accomodated.
Description: ber_printf support for O format
Fix Description: When calling ber_printf with the 'O' format, the corresponding argument must be a struct berval *. If the bval argument is NULL, nothing is written (i.e. it is safe to use a NULL here). Otherwise, bval->bv_len bytes from bval->bv_val will be written to the output.
Description: Merge sun libprldap changes on to trunk
Fix Description: Pretty much the same as the fixes made for bugs 352519 352673 on the sun_merge_branch_20060523, this just merges those changes on to the trunk. One additional change is the removal of an unused variable.
Description: Merge sun_merge_branch_20060523 changes into trunk.
Fix Description: I had to change re_encode_request to use the new ber types. Other than that, these merges are basically the sasl, ipv6, and ber types code already committed to the sun merge branch.
Description: Memleak: ldap_start_tls_s should free oidptr and dataptr
Fix Description: Just omit oidptr and dataptr and pass NULL for those arguments to ldap_extended_operation_s().
Bug: 348927
Description: Memleak: ldaptls_complete should free hostlist
Fix Description: Instead of returning directly from early error conditions, just goto the error handling code, and let that code free hostlist and do the other cleanup. I had to make sure all relevant variables were initialized to NULL in order for cleanup to work properly.
Description: The client must call ldapssl_enable_clientauth before the first operation. This means before ldap_start_tls_s in the case of startTLS. However, the certname and keypassword (if any) are stored in the connection context. ldapssl_enable_clientauth will now allocate and store this connection context if it doesn't already exist, then set the certname and keypassword.
Fix Description: Make a copy of the value parameter passed into ldap_getfirstfilter() and store it in lfd_curval. Free it in the LDAPFiltDesc free function.
Fix Description: Using the -Z flag sets the secure option, but using -ZZ (use starttls) does not. The code in several places uses a test like if (secure || isZZ) to see if it has to perform some sort of ssl/tls related initialization. The one place that was missed was before the call to PinArgRegistration() which was just if (secure).
nkinder%redhat.com
bugzilla%standard8.demon.co.uk
gerv%gerv.net
anton.bobrov%sun.com
richm%stanfordalumni.org
nhosoi%redhat.com
mozilla%weilbacher.org
bzbarsky%mit.edu
timeless%mozdev.org