(The first announcement didn't seem to go through...)
Hi,
I'm pleased to announce the release of next stable version of GnuGk.
You can download version 2.0.7 as usual from
http://www.gnugk.org/h323download.html
Currently executables are available for Linux, Windows and Solaris.
FreeBSD and Mac OS X will follow shortly.
The most notable changes are in Radius support, some virtual queue
features and many bugfixes especially on Windows.
See the attached changelog for details.
Enjoy,
Jan
Changes from 2.0.6 to 2.0.7
===========================
- RadAuth/RadAliasAuth modules can now add/remove endpoint aliases
during
endpoint registration (see radauth.txt for more info).
WARNING: GkAuthenticator Check RRQ virtual method signature has
changed!!!
- BUGFIX(GkStatus.cxx) rollback of the previous buggy patch
- fixed destination call signaling address handling (it should not
disapper
from CDRs anymore and should be consistent for auth/acct modules and
NATed
endpoints)
- various small status line improvements (command echo, up arrow repeats
the most recent command, a new debug cfg command)
- BUGFIX(gkauth.cxx) fixed call duration limit handling when multiple
authenticator modules are specified
- added utilities to the contrib dir that allow the gnugk ro run
as a Windows service process. Thanks to Franz J Ehrengruber!
- new Windows project configurations - Win32 ReleaseAcct and Win32
DebugAcct
- Windows improvements (project icon, version info, better console
handler). Thanks to Franz J Ehrengruber!
- BackSpace key is now properly handled for Windows telnet clients
- BUGFIX(RasTbl.cxx) fixed CRV handling (call reference value flag
0x8000)
- BUGFIX(RasTbl.cxx) added missing lock. Thanks to Andrey Pankov!
- BUGFIX(RasTbl.cxx) source call signaling address should not disappear
from CDRs and accounting anymore for unregistered endpoints
- BUGFIX(h323util.cxx) additional check for H225_TransportAddress
to prevent NULL pointer assertions. Thanks to Andrey Pankov!
- BUGFIX(ProxyChannel.cxx) incorrect arguments passed to FindEndpoint
could cause wrong Calling-Station-Id selection for accounting purposes
- BUGFIX(yasocket.cxx) PSocket::Select was not called when compiled
without PTRACING. Thanks to Per Kreipke!
- BUGFIX(RasTbl.cxx) fixed wrong destination info for calls, caused
by previous changes. Thanks to Alexey Livikin
- Radius Q.931 Setup authentication routines optimized.
WARNING!!!: GkAuthentication API for Q.931 checks has changed
- BUGFIX(radauth.cxx) better Framed-IP-Address detection for
unregistered
endpoints
- BUGFIX(ProxyChannel.cxx) fixed missing originating IP for some calls
from unregistered endpoints
- BUGFIX(ProxyChannel.cxx) multiple calls over a single signalling
channel
are now gracefully handled using ReleaseComplete with
newConnectionNeeded code.
Thanks to Alex Golyshev!
- BUGFIX(radauth.cxx) fixed Session-Timeout processing in RadAliasAuth
Setup check
- BUGFIX(radproto.cxx) fixed invalid signed/unsigned conversion that
caused
some functions to return unexpected results
- better Calling-Station-Id handling for acct/auth modules
- BUGFIX(RasSrv.cxx) ACF will now contain rewritten destinationInfo
(by a virtual queue) only if the matching ARQ comes from the virtual
queue
- better Called-Station-Id handling - now it is set to an alias that
actually matched the destination endpoint
- BUGFIX(WaitingARQ.cxx) better mulithreading safety
- BUGFIX(RasSrv.cxx) LCF messages contain now correct destinationInfo
- e.g. for gateways LCF will contain dialed number, not the gateway
alias
- gateway and neighbor prefixes match can also occur with alias types
(in addition to dialedDigits) partyNumber and h323_ID (containing only
0-9#*)
- added new VirtualQueuePrefixes and VirtualQueueRegex config variables
to [CTI::Agents] section. These make possible to call virtual queue
not only with the exact alias name, but also with an alias that
matches
configured prefixes or configured regular expression. Thanks to Max
Speransky
- renamed config variable VirtualQueue to VirtualQueueAliases
and CTI_Timeout to RequestTimeout (backward compatibility is
maintained)
- BUGFIX(RasSrv.cxx) crypto tokens are not copied from RAS requests to
RAS
replies, as this does not make any sense in the current form. H.235
Auth
Procedure I should be implemented instead and crypto tokens should be
generated
by the gatekeeper, not just copied
- BUGFIX(gkauth.gxx) fixed segfault in H.235 capability selection
routine. Thanks to Daniel Liu
- changed handling of multiple accounting modules (added new
"alternative" control)
- BUGFIX(RasTbl.cxx) fixed connect/disconnect time handling and call
duration
calculation. Thanks to Aivis Olsteins
- BUGFIX(ProxyChannel.cxx) acctstart events can now correctly
log an IP address for unregistered endpoints
- BUGFIX(RasTbl.cxx) make sure format of ACXF message generated by
printcurrentcalls matches the format of normal ACFs
- BUGFIX(ProxyChannel.cxx) Q.931 setup authentication now correctly
updates number of active/total calls for unregistered endpoints
and records unconnected CDRs for rejected calls due to auth failure
- BUGFIX(ProxyChannel.cxx) if Home is specified in the config, RTP proxy
sockets will bind to this specific address - this should solve
problems
with proxy gatekeepers running on the same machine. Thanks to Julius
Stavaris
- accounting updates for calls in progress. Currenly only RadAcct
module implements this type of accounting event
- BUFGIX(radproto.cxx) socket write operation is atomic now
- BUGFIX(RasSrv.cxx): ARQ was not sent from child GK to parent GK
for the second and subsequent calls to the same OZEP
- BUGFIX(radproto.cxx): unaligned 32-bit access crashed GK on systems
with Alignment Check enabled
- improved Radius h323-xxx attributes handling (now understands
attributes both with embedded name strings and without them)
- BUGFIX(radauth.cxx): Session-Timeout RADIUS attribute was processed
as a string-type attribute, not integer-type attribute
- BUGFIX(radauth.cxx): Q.931 Setup authentication stopped to work due
to previous changes
- BUGFIX(RasTbl.h): better call duration limit accuracy (was always 1
second longer)
- BUGFIX(singleton.cxx): incorrect singleton instance counting with
different trace levels
- BUGFIX(Makefile): GK_NOANSI moved out of LDAP section
--
Jan Willamowius, jan@willamowius.de, http://www.willamowius.de/
-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn
_______________________________________________
List: Openh323gk-users@lists.sourceforge.net
Archive: http://sourceforge.net/mailarchive/forum.php?forum_id=8549
Homepage: http://www.gnugk.org/
