mirror of
https://github.com/asterisk/asterisk.git
synced 2025-10-31 02:37:10 +00:00
d605a00c13
https://origsvn.digium.com/svn/asterisk/trunk ........ r222110 | kpfleming | 2009-10-05 14:45:00 -0500 (Mon, 05 Oct 2009) | 25 lines Allow non-compliant T.38 endpoints to be supportable via configuration option. Many T.38 endpoints incorrectly send the maximum IFP frame size they can accept as the T38FaxMaxDatagram value in their SDP, when in fact this value is supposed to be the maximum UDPTL payload size (datagram size) they can accept. If the value they supply is small enough (a commonly supplied value is '72'), T.38 UDPTL transmissions will likely fail completely because the UDPTL packets will not have enough room for a primary IFP frame and the redundancy used for error correction. If this occurs, the Asterisk UDPTL stack will emit log messages warning that data loss may occur, and that the value may need to be overridden. This patch extends the 't38pt_udptl' configuration option in sip.conf to allow the administrator to override the value supplied by the remote endpoint and supply a value that allows T.38 FAX transmissions to be successful with that endpoint. In addition, in any SIP call where the override takes effect, a debug message will be printed to that effect. This patch also removes the T38FaxMaxDatagram configuration option from udptl.conf.sample, since it has not actually had any effect for a number of releases. In addition, this patch cleans up the T.38 documentation in sip.conf.sample (which incorrectly documented that T.38 support was passthrough only). (issue #15586) Reported by: globalnetinc ........ git-svn-id: https://origsvn.digium.com/svn/asterisk/branches/1.6.1@222112 65c4cc65-6c06-0410-ace0-fbb531ad65f3
140 lines
6.9 KiB
Plaintext
140 lines
6.9 KiB
Plaintext
===========================================================
|
|
===
|
|
=== Information for upgrading between Asterisk 1.6 versions
|
|
===
|
|
=== These files document all the changes that MUST be taken
|
|
=== into account when upgrading between the Asterisk
|
|
=== versions listed below. These changes may require that
|
|
=== you modify your configuration files, dialplan or (in
|
|
=== some cases) source code if you have your own Asterisk
|
|
=== modules or patches. These files also includes advance
|
|
=== notice of any functionality that has been marked as
|
|
=== 'deprecated' and may be removed in a future release,
|
|
=== along with the suggested replacement functionality.
|
|
===
|
|
=== UPGRADE-1.2.txt -- Upgrade info for 1.0 to 1.2
|
|
=== UPGRADE-1.4.txt -- Upgrade info for 1.2 to 1.4
|
|
=== UPGRADE-1.6.txt -- Upgrade info for 1.4 to 1.6
|
|
===
|
|
===========================================================
|
|
|
|
As of 1.6.1.7:
|
|
|
|
* The firmware for the IAXy has been removed from Asterisk. It can be
|
|
downloaded from http://downloads.digium.com/pub/iaxy/. To have Asterisk
|
|
install the firmware into its proper location, place the firmware in the
|
|
contrib/firmware/iax/ directory in the Asterisk source tree before running
|
|
"make install".
|
|
|
|
* T.38 FAX error correction mode can no longer be configured in udptl.conf;
|
|
instead, it is configured on a per-peer (or global) basis in sip.conf, with
|
|
the same default as was present in udptl.conf.sample.
|
|
|
|
* T.38 FAX maximum datagram size can no longer be configured in updtl.conf;
|
|
instead, it is either supplied by the application servicing the T.38 channel
|
|
(for a FAX send or receive) or calculated from the bridged endpoint's
|
|
maximum datagram size (for a T.38 FAX passthrough call). In addition, sip.conf
|
|
allows for overriding the value supplied by a remote endpoint, which is useful
|
|
when T.38 connections are made to gateways that supply incorrectly-calculated
|
|
maximum datagram sizes.
|
|
|
|
As of 1.6.1.6:
|
|
|
|
* There have been some changes to the IAX2 protocol to address the security
|
|
concerns documented in the security advisory AST-2009-006. Please see the
|
|
IAX2 security document, doc/IAX2-security.pdf, for information regarding
|
|
backwards compatibility with versions of Asterisk that do not contain these
|
|
changes to IAX2.
|
|
|
|
From 1.6.1.1 to 1.6.1.2:
|
|
|
|
* Beginning with this release, Asterisk's internal methods of
|
|
negotiating T.38 (FAX over IP) sessions changed in
|
|
non-backwards-compatible ways. Any applications that previously used
|
|
AST_CONTROL_T38 control frames will have to be upgraded to use
|
|
AST_CONTROL_T38_PARAMETERS control frames instead; app_fax.c is a good
|
|
example of how to generate and respond to these frames. These changes
|
|
were made to solve significant T.38 interoperability problems between
|
|
Asterisk and various SIP/T.38 endpoints identified by many users of
|
|
Asterisk.
|
|
|
|
From 1.6.0.1 to 1.6.1:
|
|
|
|
* The ast_agi_register_multiple() and ast_agi_unregister_multiple()
|
|
API calls were added in 1.6.0, so that modules that provide multiple
|
|
AGI commands could register/unregister them all with a single
|
|
step. However, these API calls were not implemented properly, and did
|
|
not allow the caller to know whether registration or unregistration
|
|
succeeded or failed. They have been redefined to now return success
|
|
or failure, but this means any code using these functions will need
|
|
be recompiled after upgrading to a version of Asterisk containing
|
|
these changes. In addition, the source code using these functions
|
|
should be reviewed to ensure it can properly react to failure
|
|
of registration or unregistration of its API commands.
|
|
|
|
* The ast_agi_fdprintf() API call has been renamed to ast_agi_send()
|
|
to better match what it really does, and the argument order has been
|
|
changed to be consistent with other API calls that perform similar
|
|
operations.
|
|
|
|
From 1.6.0.x to 1.6.1:
|
|
|
|
* In previous versions of Asterisk, due to the way objects were arranged in
|
|
memory by chan_sip, the order of entries in sip.conf could be adjusted to
|
|
control the behavior of matching against peers and users. The way objects
|
|
are managed has been significantly changed for reasons involving performance
|
|
and stability. A side effect of these changes is that the order of entries
|
|
in sip.conf can no longer be relied upon to control behavior.
|
|
|
|
* The following core commands dealing with dialplan have been deprecated: 'core
|
|
show globals', 'core set global' and 'core set chanvar'. Use the equivalent
|
|
'dialplan show globals', 'dialplan set global' and 'dialplan set chanvar'
|
|
instead.
|
|
|
|
* In the dialplan expression parser, the logical value of spaces
|
|
immediately preceding a standalone 0 previously evaluated to
|
|
true. It now evaluates to false. This has confused a good many
|
|
people in the past (typically because they failed to realize the
|
|
space had any significance). Since this violates the Principle of
|
|
Least Surprise, it has been changed.
|
|
|
|
* While app_directory has always relied on having a voicemail.conf or users.conf file
|
|
correctly set up, it now is dependent on app_voicemail being compiled as well.
|
|
|
|
* SIP: All of the functionality in SIPCHANINFO() has been implemented in CHANNEL(),
|
|
and you should start using that function instead for retrieving information about
|
|
the channel in a technology-agnostic way.
|
|
|
|
* If you have any third party modules which use a config file variable whose
|
|
name ends in a '+', please note that the append capability added to this
|
|
version may now conflict with that variable naming scheme. An easy
|
|
workaround is to ensure that a space occurs between the '+' and the '=',
|
|
to differentiate your variable from the append operator. This potential
|
|
conflict is unlikely, but is documented here to be thorough.
|
|
|
|
* The "Join" event from app_queue now uses the CallerIDNum header instead of
|
|
the CallerID header to indicate the CallerID number.
|
|
|
|
* Support for Taiwanese was incorrectly supported with the "tw" language code.
|
|
In reality, the "tw" language code is reserved for the Twi language, native
|
|
to Ghana. If you were previously using the "tw" language code, you should
|
|
switch to using either "zh" (for Mandarin Chinese) or "zh_TW" for Taiwan
|
|
specific localizations. Additionally, "mx" should be changed to "es_MX",
|
|
Georgian was incorrectly specified as "ge" but should be "ka", and Czech is
|
|
"cs", not "cz".
|
|
|
|
* If you use ODBC storage for voicemail, there is a new field called "flag"
|
|
which should be a char(8) or larger. This field specifies whether or not a
|
|
message has been designated to be "Urgent", "PRIORITY", or not.
|
|
|
|
From 1.6.1.1 to 1.6.1.2:
|
|
|
|
* Support for Taiwanese was incorrectly supported with the "tw" language code.
|
|
In reality, the "tw" language code is reserved for the Twi language, native
|
|
to Ghana. If you were previously using the "tw" language code, you should
|
|
switch to using either "zh" (for Mandarin Chinese) or "zh_TW" for Taiwan
|
|
specific localizations. Additionally, "mx" should be changed to "es_MX",
|
|
Georgian was incorrectly specified as "ge" but should be "ka", and Czech is
|
|
"cs", not "cz".
|
|
|