4.3.62

Changes for 4.3.62

Changes to branch 4.3 after version 4.3.61.

Commits

• KZOO-15: Ensure min value of max_recording_time_limit - by ming luo

  max_recording_time_limit can be reset by values from system_config/callflow by kazoo_media_maintenance:migrate(). To prevent reset again, the original callflow (or any originals) data shall be removed after the migration.

  It's also required that the system-config default should be 10800, as seconds in 3 hours.

4.3.61

Changes for 4.3.61

Changes to branch 4.3 after version 4.3.60.

Commits

• update ledger summary normalizer - by icehess

  Fix normalizer for ledgers summary API so it extract account ID from view's key correctly.

  In previous PR for PRDO-30 the ledgers account summary was change to have account_id, timestamp and doc._id as view's compound key, but the normalizer in crossbar module didn't updated accordingly.

  master: #6127

4.3.60

Changes for 4.3.60

Changes to branch 4.3 after version 4.3.59.

Commits

• HELP-10868: ignore channel bridge and log at info when a limit is exceeded - by James Aimonetti

• MCCP-206: insert recv_dtmf immediately - by mark2600

  • queueing this will cause it to not take effect immediately

• HELP-10392: Honor ring group members set delay (#6045) - by Harenson Henao

  Master: #6045

  When an endpoint appears multiple times in a ring group, treat them as

  separate endpoints according to their configured delay and timeout

  parameters. Also, sort the resolved endpoints by delay before sending

  them to ecallmgr.

• KZOO-20/21: fix app used_by for port_in number - by Ming

  There are a few bugs addressed in this PR.

  Remove used_by json object under a port_in number when the associated callflow is deleted.

  An earlier bug caused used_by not set when a port_in number is

  in rejected state. Although this has been fix in this PR,

  the ramification of earlier numbers without this fix still exists.

  Therefore, this commit checks if used_by is not set properly when

  the port_in transitions to completed state.

  It also fixes another bug that a newly created phone_number as a result of

  port_in completed will preserve used_by app. That has been missing.

4.3.59

Changes for (4.3.59)

Changes to branch 4.3 after version 4.3.58.

Commits

• remove dth from the kazoo-applications metapackage for subsequent releases - by Sean Wysor

• Pusher and ACDC should be related not required - by Sean Wysor

• fix audit_summary view - by icehess

  services/day_summary_by_date view is erroring out with **"reduce is not

  shrinking fast enough"** error in BigCouch (only in Bigcouch, CouchDB 2 is fine so far!).

  This is changing the view so it exactly like services/day_summary_by_source

  but the view's key is swap, it is date string and source service.

  We instead reshaping the result in crossbar.

  Also export kz_util:log_stacktrace/3

  master: #6115

• More circleci updates - by icehess

  This is the latest fixes for application's CircleCI config file:

  We found out the CHANGED variable are not passed to some of the targets, so they won't actually runs against the the app report and no won't report any error/warnings at all.

  • fix check_unstaged step show it shows the changes in the app folder

  • fix dialyzer so it actually dialyzes the changed files in the app folder

  • fix code_checks so it runs on changed files in the app folder

  • fix fmt

  • add coverage report step (I think it needs per repo activation in coveralls.com site)

  • UPDATE: disabling coverage report. It seems we need to set a token when using a CI other than than Travis-CI. Also it is crashing right now which we need to investigate it later. And we need to find out what to do with other repos too.

  • fix deps.mk so it adds coveralls as dependency in circleci env

  • Use tee as much as possible and print important variables

  master: #6109

• PROD-30: some enhancements to run custom tasks + template - by icehess

  This is some changes required for PROD-30, mostly related to template rendering and tasks.

  • Move trigger-able macros to kazoo task core app

  • Add option to kz_json:from_list_recursive/2 to hint that don't treat list as ASCII chars list (printable chars) so empty list won't be converted to binary instead of a genuine empty list

  • Add new type kz_time:kazoo_month() to refer to {Year::kz_time:year(), Month::kz_time:month}

  • Add kz_template:render_with_option/3 to pass render time options like translation_fun

  • Some enhancements to ledgers and service audit view to get summary quicker

  Backport of #6070

• fix publish_usurp_publisher key - by lazedo

  This is a backport of @lazedo 's change in dc723cf

• HELP-10397 Use default value for type in group pickup - by ming luo

  Previously type was not required; if not included the action would

  essentially be skipped. The schema was then updated to require type

  but existing feature code callflows still had empty data objects

  with no type defined - thus still "not working".

  Default the type to extension as that is the common type used by

  SmartPBX-configured feature codes.

• fix check-dialyzer script to properly fix full path - by Hesaam Farhang

  For kazoo apps repo, we use the full path as changed file, and pass that to check_dialyzer script. But that script is not properly converting the erl file path (full path) to correct ebin file (dropping the leading "/" from the beginning) causing to not dialyze anything at all.

  master: #6102

• gcm accepts kz_json object not map - by lazedo

  unlike firebase that accepts map

• removing travis-ci - by icehess

  We're going to use only CircleCI from now on. Deleting the Travis-CI config file.

  • add coverage report step

  • add back make fmt step

  • some cleanup in cicrleci config and better step's descriptions

  • update couchdb docker image to the latest stable release 2.3.0

  This PR probably needs the travis app to be disabled in Github repo settings to pass the CI requirement.

  master: #6096

4.3.58

Changes for (4.3.58)

Changes to branch 4.3 after version 4.3.57.

Commits

• KZOO-8: correctly set switch_nodename for kapps_call on record_stop - by Mark Magnusson

  • the code used to send the AMQP message to store the recording uses

  the switch_nodename to target that message, set it correctly to fix the

  issue of recordings not storing properly when it is not set

• HELP-10722: clarify when to build failover - by James Aimonetti

  Build failover endpoint if endpoint is unregistered and failover is

  present on the endpoint's bridge_endpoint record.

4.3.57

Changes for (4.3.57)

Changes to branch 4.3 after version 4.3.56.

Commits

• KCRO-1 : support a request parameter details on the available service plans - by bitbashing

  Adds an optional parameter to the services available API which when true will include the full plan in the listing.

  {{api_server}}/accounts/{{account_id}}/services/available?details=true

  Master PR #6085

• make compile-lean added to makefiles - by swysor

4.3.56

Changes for (4.3.56)

Changes to branch 4.3 after version 4.3.55.

Commits

• Updated CircleCI config and integration with build system - by Sean Wysor

• Fixed compile test in kazoo_documents - by Sean Wysor

• KAZOO-6122: when fetching the account id by IP, return an error if an object is disabled - by karl anderson

  In the case of a valid document, that is enabled, and system_config set to use sip_auth, AND the associated user or account disabled kapps_util will return {ok, []}

  ie: ok with no additional parameters to add to the CCVs reg_route_req assumed (happy case) that if it was found an account id was associated. However, in this case no account id is proved leading to a route_req replay with no account id and a loop.

  From an admin perspective, setting up an account with ip auth, disabling the device/user/account associated and then calling the number results in a significant spike in CPU and/or memory.

• add type required for group_pickup_feature - by ming luo

  1. Add type is required for group_pickup_feature's callflow json

  2. Change b_play to b_prompt

4.3.55

kazoo Changes

Changes to 4.3 after version 4.3.54

Unticketed Commits

• [4.3] correct get_values_and_keys return when duplicate keys exist (#6035) also test encoding/decoding query strings more thoroughly by James Aimonetti
• Allow generic branching of callflow based on capture group (#6032) Most callflow actions that include a feature code option have a separate module for handling the capture group portion, and a main action module for the feature. Rather than require all actions wanting to be feature-code-able to create a corresponding feature code module, when the action only wants to use the capture group to re-hunt and branch to the found callflow, this callflow action branch_bnumber can be put at the end to perform the hunt. by James Aimonetti
• [4.3] HELP-10194: Add utc_unix_timestamp_ms key to pushed objects via pusher (#6031) by Harenson Henao

4.3.54: [4.3] HELP-10237: refactor decision-making for call recording (#6026)

kazoo Changes

Changes to 4.3 after version 4.3.53

Unticketed Commits

• [4.3] HELP-10237: refactor decision-making for call recording (#6026) * [4.3] HELP-10237: refactor decision-making for call recording Added tests for kz_endpoint when merging call recording settings to clarify the behaviour. Updated documentation to clarify precedence and the duality of account and endpoint call recording settings. by James Aimonetti

4.3.53: add logging in gen_listener and don't shortcut shutdown (#6023)

kazoo Changes

Changes to 4.3 after version 4.3.52

Unticketed Commits

• add logging in gen_listener and don't shortcut shutdown (#6023) * ecallmgr node sup which is a supervisor under a supervisor (fs_sup) is terminated with shutdown. if we find a way to properly shutdown the supervisor with a reason we can re-add the shutdown shortcut (cherry picked from commit 509a03ebac5527e280f81c6af6221d9eafc958af) by lazedo
• HELP-45446: [4.3] move when termination handlers are running to cf_exe termination (#5980) * HELP-45446: adding hangup cause and disposition to command error event Adding the hangup reason to command execution error event can help the caller to determine if it need to continue the processing or not. For example cf_device can check if the bridge command failed because of hangup and if yes, then stop the callflow right there, so cf_exe won't have any race condition of continuing to the next callflow and receiving channel_destroy event. * check if channel is destory before relay the message in cf_exe instead * make sure hangup cause is defined in error event * send hangup command if it did not sent already by icehess
• [4.3] sort user in directory consistently (#6015) by Ming Luo