compiled using orbit-idl, which uses the C preprocessor to do its work, and
the C preprocessor has it's arguments transformed (correctly) by
buildlink2.
- GDL handling has been totally rewritten and is now more clean.
- It is now possible to destroy GDL "rename at end" entry like any other one.
- In the GDL list, sizes are now displayed using locale format.
- search result handling has been partially rewritten and is now more clean.
- The search result size is now displayed using locale format.
- old style user list is no more supported, only new one is (the one using
list cache).
- user share lists handling has been partially rewritten and is now more clean.
- The user share lists are now displayed using locale format.
- a (new) buggy clone seems to send string containing illegal character ('\0').
Because this character is also the C string end character, it truncates string
and DCTC crashs because an impossible case occurs. This version avoids the
problem.
headers in ${BUILDLINK_DIR}, simply create BUILDLINK_CPPFLAGS.<pkg>
variables whose values are appended to CPPFLAGS, which are automatically
passed to the configure and build processes.
BUILDLINK_TRANSFORM.<pkg> has little use in buildlink2 since packages are
now told that the software may be found where it really is installed, not
in ${BUILDLINK_DIR} as was the case with buildlink1. Eventually, these
variables will be declared unsupported by buildlink2.
extension Makefile fragments, because they really don't have anything to
do with the buildlink[12] frameworks. Change all the Makefiles that use
application.buildlink.mk and extension.buildlink.mk to use application.mk
and extension.mk instead.
- adapt for NetBSD 1.6 and later which needs an "ifconfig stf0 create"
(also needed on FreeBSD).
Patch contributed by Kimura Fuyuki <fuyuki@hadaly.org> in private mail.
* Added version and length sanity checks for incoming messages
* Fixed possible SEGV when logging DA traffic
* Added patch from Arun Barboza that makes attribute compares case insensitive
as per RFC 2608
* Added patch from Garth Bushell that fixes an SLP url parsing problem.
* Added patch from Arun Barboza to handle '==' SLPv1 predicates. This patch
allows better interoperability with Novell SLP networks.
* Added fix from Arun Barboza to make V1 DAAdverts use the same encoding as
the request
* Fixed a problem in SLPProcessMessage where errorcodes were not being returned
resulting in attempts to send non-allocated sendbufs (i.e. SEGV).
* Fixed a problem where SLPReg() would hang if used against a DA or SA that
prematurely dropped connections
* Force use of numeric IP addresses in all DAAdvert and SAAdvert messages.
* Fixed bug where idle connections to DAs were not being aged.
* Added code so that 'slptool -v' will return the path of the slp.conf file
libslp.so is configured to use.
* mod_quota removed due to lack of active maintenance
* New version of mod_ldap (2.8.10)
* New directives added: Define, <IfDefine>, <IfModule>, TimeoutSession
* Directives removed: AllowChmod, ShowDotFiles
* STOU support added
* LFS support added
* Fixed various minor bugs
- The editor could erroneously allocate dynamic instead of static memory.
- Fixed a spurious pause that would occur if the first callout started after a
cold boot had a delay of 0.
- Fixed a problem with repeatedly adding and removing callouts in atomically
executed code.
- Fixed several bugs in error handling.
- A better fix for errors in atomic code, in combination with multi-level
catches.
- Fixed a config file reading bug that could occur if the server was restarted.
- Fixed a bug in saving/restoring variables in an object that privately
inherited another.
- Prevent code from being erroneously optimized away in rare cases.
- Fixed a bug in handling some regular expressions.
- Fixed several bugs which could, in rare cases, cause nil to turn up in
mappings that had been accessed from atomically executed code.
- Fixed a bug in error recovery for compiling switch statements.
The update was contributed by Scott Thompson in private e-mail.
Changes:
Import the patch from Robert Elz:
1) do not reset the interface flags when terminating .
2) do not output to default logname when terminating,
if -f option is not given.
