CTArchitect 10.4.5
Version 10.4.5.8 (24/08/2017)
- Fixed failure to detect Hasp Key during Remote Desktop
- Updated the way
INFOrequests are handled by CTArchitect. Some SIP-trunks useINFOrequests as a "ping" to check if the call is still active. This is a bad practice sinceSession-Timers/ re-INVITE/UPDATEare meant for this, as explained in RFC 4028 (implemented in CTArchitect 10.4.5). But since Dialogic puts theINFOmethod in theAllowheader, it is not prohibited. From now on, CTArchitect will respond to theINFOrequests with a200response (instead of ignoring theINFOrequest).
Version 10.4.5.7 (25/07/2017) (Happy Birthday Louise Brown)
- Fix in order to keep question record info in case of reuse voice channel after previous record
- Rewrite Fix to prevent race condition on autdia mandia
- Fixed failure to detect Hasp Key during Remote Desktop
Version 10.4.5.6 (05/07/2017)
- Fixed an issue with filenames used in plays/recordings when it contains ASCII characters between 128 and 255 (e.g.
é). The file was not found, and the logging contained incomplete/merged lines, and showed the character as a?. - Detect incorrect flowsettings at startup
Version 10.4.5.5 (30/05/2017)
- Fixed crash during Supervisor logon when CTArchitect is not completely started yet.
- Fixed possible crash when using OutboundCall that is already disconnected (e.g. when Agent status changes). Zendesk #32276.
- Fixed a setup problem with configuration tables when upgrading from 5.3.x to 5.4.5+. The existing TXT configuration tables
were not upgraded because the initial CSV configuration tables were installed. It only concerns these configuration tables:
Carrier,Country,DIDandIP. As a workaround, the TXT files could be imported in CTArchitect, or the CSV files can be removed before (re)starting CTArchitect. From now on, the initial CSV configuration tables are not installed whencountry.txtis present in the Project directory, which will cause CTArchitect to correctly import the existing TXT configuration tables.
Version 10.4.5.4 (20/04/2017)
- Fixed problem in setups while comparing versions which could cause e.g. v10 to be considered older than v2.7. This fix applies to CTArchitect, CTNode, GlobalCCA and WebProd setups, and was first seen on MDAC prerequisite check.
- Fixed some problems with Microsoft Visual C++ Redistributable in setups:
- When installing Microsoft Visual C++ Redistributable, in most cases multiple versions within the same track were installed. E.g. when installing Microsoft Visual C++ 2015 v14.0.24215, v14.0.24212 was also installed. This bug applies to Microsoft Visual C++ 2008 up till 2015. Except for the unneeded installation, there is no impact.
- Updated version-check of Microsoft Visual C++ 2015 Redistributable (both 32-bit and 64-bit), according to a new recommendation of Microsoft. This fixes problems when running the installer on systems where Microsoft Visual C++ 2017 Redistributable (14.10.x) or some versions of TFS testing framework are already installed. An error message box is displayed (errorcode 0x666 or 0x80070666), after which the setup continuous.
Version 10.4.5.3 (15/03/2017)
- Fixed startup problems with CTWatch because the packaged Visual C++ runtime is 2015 (vc14) and CTWatch was built with Visual Studio 2013 (vc12). CTWatch is now also built with Visual Studio 2015 (vc14). This is a fake release of CTArchitect to allow identification of the new setup containing the updated CTWatch. CTArchitect 10.4.5.3 and 10.4.5.2 are identical.
Version 10.4.5.2 (08/03/2017)
- Fixed a problem when updating configuration tables from CTArchitect in GUI-mode connected to a CTArchitect in service-mode, when the original configuration file didn't exist.
- Fixed possible corrupt data (or even crashes) on CCA when sending statistics. The maximum number of statistics being sent in a batch is reduced from 100 to 50.
- Fixed problem with reading
stat<n>.cfgfiles, when the line contains more than 254 characters. The maximum limit is now increased to 1023 characters. In a later version, further improvements will be done (Unicode support, further robustness).
Version 10.4.5.1 (09/02/2017)
-
Added support for SIP Session timers (RFC 4028).
-
Session timers are an extension to SIP:
- they prevent infinite calls (e.g. after a server is restarted without closing existing sessions first).
- they keep firewall hole punching "open".
-
Current implementation:
- the minimum expiration timeout is 90 seconds. Smaller expiration intervals are rejected with a response code 422 "Session Interval Too Small".
- the proposed default expiration timeout is 1800 seconds.
- the refreshing is done by sending re-INVITEs. For incoming requests, both UPDATEs and re-INVITEs are supported.
- the refresher is choosen by the UAS (the party receiving the INVITE). When HMP is UAS, it chooses the refresher as the one proposed by the UAC (the party sending the INVITE), or himself if no refresher was proposed. When HMP is UAC, it is not proposing a refresher.
-
This feature is disabled by default, and can be enabled by setting this registry key to 1 (
DWORD):HKEY_LOCAL_MACHINE\SOFTWARE\WOW6432Node\MI4C\CTArchitect\HMP1_SessionTimersEnabled -
Even when remote side doesn't actively support Session Timers, this feature can be useful: when the remote side doesn't support Session Timers, the local side (HMP) will be the refresher, using the default expiration timeout. The remote party will respond to the re-INVITEs as specified by SIP (RFC 3261).
-
Possible future development is the configuration of: default refresh interval, minimum refresh interval, refresher selection preference, ...
-
Further details can be found in topic "SIP Session Timer" in HMP 3.0 Release Update .
-
-
Added original SQL error in message box when connection to Global database fails.
-
Added possibility to retry when DSN to CCA database could not be opened.
-
Updated confusing message in logging about collisions when a gc_MakeCall/TaskFail error happens. Collision is only one of the possible reasons.
-
Updated Setups to automatically use SIP instead of H.323 on new installations.
This means:- The default Route table has been updated to use SIP instead of H.323 for outbound calls to Agents.
- The default value of the CTArchitect setting
SoftPhone Use H.323is updated fromYestoNo.
-
Fixed crash when CTNode is passing an empty parameter or extension block to CTArchitect.
-
Fixed race condition during Dial (
AUTDIA), ManualDial (MANDIA), LocationManualDial (LMANDI) and BlindTransfer (BLITRA). Several states were executed twice, which resulted in failure to call the number. This error was present in the logging:Internal Number <#> could not be found in CCA with ID <#>(or other errors regarding internal numbers). -
Fixed serveral issues in setups:
- Some folder locations were wrong when using the "directory suffix" feature.
- Location of CTDesign was wrong when registering as service.
- Startup parameters of CTArchitect were wrong when starting as service.
- Startup icons were created in root Start menu folder instead of
Programs\StartUpsubfolder when enabling both "autologon" and "autostart". (Fixed in both CTArchitect and CCA). - AutoLogon wizard page was missing (and consequently, AutoStart icons were not created).
-
Fixed argument parsing to allow quoted arguments containing spaces. This fixes an issue with
-workingdir:<dir>, when<dir>contains spaces. In this case you should put the complete argument between double quotes:"-workingdir:<dir>". -
Fixed crash during startup in service mode, when an unknown startup option was specified. The resulting MessageBox "Unknown Option" that is normally transferred to the CTArchitect in GUI-mode caused the crash because CTArchitect is not initialized yet. The MessageBox is now assumed "acknowledged" without waiting for the result, and CTArchitect will close immediately. Since no logging is possible this early in CTArchitect initialization, this solution is still sub-optimal.
-
Fixed an inconsistency in the statistics CLI. The CLI value depended on wether the call between Cati/Supervisor and CTArchitect was already established. From now on, the CLI always contains the AgentName (without spaces). Fixes TFS #3282.
-
Fixed wrong disconnect cause was used on all calls that were taken back. The disconnect cause of the first disconnected call was used for all calls. Fixes TFS #3293.
-
Refactored various internal lists and maps to simplify housekeeping of calls, agents and other collections.
-
Refactored code to clean up old technologies and functionality. Isolate BTrieve logic.
-
All fixes up to 10.4.4.2.