#acl PaulHowarth:read,write,admin,revert,delete All:read === Monday 18th May 2015 === ==== Fedora Project ==== * Updated `perl-Net-SSLeay` (1.68) in Rawhide to revert upstream's dropping of `SSLv3_method` with OpenSSL 1.0.2 ([[RedHatBugzilla:1222521|Bug #1222521]]) ==== Local Packages ==== * Updated `dovecot`: . Updated `dovecot` to 2.2.18: * `director`: Login UNIX sockets were normally detected as `doveadm` or `director` ring sockets, causing it to break in existing installations * `sdbox`: When copying a mail in alt storage, place the destination to alt storage as well . Updated `pigeonhole` to 0.4.8: * LDA `Sieve` plugin: Dovecot changed the `deliver_log_format` setting to include `%{delivery_time}`; this prompted changes in Pigeonhole that make this release dependent on Dovecot ≥ v2.2.17 * Implemented magic to make `sieve_default` script visible from !ManageSieve under a configurable name, so users can see the default rules, edit them and store a private adjusted version; this could also be achieved by copying the default script into the user's script storage, but updates to the global `sieve_default` script would be ignored that way * !ManageSieve: Implemented support for reporting command statistics at disconnect; statistics include the number of bytes and scripts uploaded/downloaded/checked and the number of scripts deleted/renamed * Fixed problem in address test: erroneously decoded mime-encoded words in address headers * `extprograms` plugin: Fixed failure occurring when connecting to script service without the need to read back the output from the external program * Fixed bug in script storage path normalization occurring with relative symbolic links below root * Fixed and updated various parts of the documentation * !ManageSieve: Used "`managesieve`" rather than "`sieve`" as login service name, which means that all `managesieve`-specific settings were ignored * !ManageSieve: Storage quota was not always enforced properly for scripts uploaded as quoted string; nobody uses that, but it is allowed in the specification and we support it, so it should work properly * Updated `perl-Capture-Tiny` to 0.30: * Fix double filehandle close error with tee on Windows (which started warning during the `perl` 5.21.x series, causing tests to fail) * Updated `perl-Module-Build` to 0.4212: * Revert "Stop using version->normal(); prefer stringify()" * Updated `perl-Net-SSLeay` as per the Fedora version * Updated `perl-Search-Elasticsearch` to 1.20: . Deprecated: * `Search::Elasticsearch::Client::Direct` in favour of `Search::Elasticsearch::Client::1_0::Direct` . New features: * Added support for structured JSON exceptions in `Elasticsearch` 2.0 * Added support for plugins * Added `Search::Elasticsearch::Client::2_0::Direct` for the upcoming `Elasticsearch` 2.0 with these changes: * Removed `delete_by_query()` * Removed `termvector()` * Removed `indices.delete_mapping()` * Removed `nodes.shutdown()` * Removed `indices.status()` * Added `terminate_after` param to `search()` * Added `dfs` param to `termvectors()` * Removed `filter_keys` param from `indices.clear_cache()` * Removed `full` param from `indices.flush()` * Removed `force` param from `indices.optimize()` * Removed `replication` param from all CRUD methods * Removed `mlt()` method . Bug fix: * The bulk buffer was being cleared on a `NoNodes` exception . Added methods: * `field_stats()` . Added params: * `allow_no_indices`, `expand_wildcards`, `ignore_unavailable` to `cluster.state()` * `fielddata_fields` to `search()` * `master_timeout` to `indices.get_template()` and `indices.exists_template()` * `detect_noop` to `update()` * `only_ancient_segments` to `upgrade()` * `analyze_wildcards`, `analyzer`, `default_operator`, `df`, `lenient`, `lowercase_expanded_terms`, and `q` to `count()`, `search_exists()` and `indices.validate_query()` . Removed methods: * `benchmark.*` - never released in `Elasticsearch` . Also: * Arrays of `enum` query string params are now flattened as CSV * `enum expand_wildcards` also accepts: `none`, `all` * `Search::Elasticsearch` is no longer a `Moo` class * Updated `elasticsearch.org` URLs to use `elastic.co` instead * The request body is retained in exceptions * Upgraded `Hijk` dependency to 0.20 ----