[Rt-commit] rt annotated tag, rt-4.2.0rc1, created. rt-4.2.0rc1

Alex Vandiver alexmv at bestpractical.com
Thu Sep 5 17:16:14 EDT 2013


The annotated tag, rt-4.2.0rc1 has been created
        at  195eedbe2a4e93625813feaad065356d213ebfec (tag)
   tagging  80b999bdc5d1f109506e868feb4cd97aeb573bb5 (commit)
  replaces  rt-4.1.23
 tagged by  Alex Vandiver
        on  Thu Sep 5 16:35:33 2013 -0400

- Log -----------------------------------------------------------------
version 4.2.0rc1
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)

iEYEABECAAYFAlIo6xUACgkQMflWJZZAbqCH3gCfXOXkLm9N4uG6TSGyv3OXxgzJ
7xkAoMBxGAfs2pNK1PtDraCN+bFtRHIP
=1e5n
-----END PGP SIGNATURE-----

Alex Vandiver (172):
      Dependency walker
      Walk large collections in batches of 100 results
      Add the concept of a UID to records
      Base-bones Dependencies method, which examines _Accessible for two common fields
      Attributes on objects
      Transactions, optimized to object types that have them
      ObjectCustomFieldValues, optimized to object types that have them
      Ensure that RT::Link's ClassAccessible is built early enough
      Dependencies
      Ensure that we always load merged tickets separately
      Bare-bones serializer
      Simplest Serialize method possible
      Use _ClassAccessible and ->can to deduce possible inflated columns
      Serialize ObjectType and ObjectId into a single UID reference
      Slightly more comprehensive set of base objects to copy
      Output serialized data into bite-sized chunks
      Serializing less than the whole db
      Serialize some global properties all of the time
      Better Serialize methods
      Allow us to skip deleted tickets
      Prepend groups and queues with the organization we were serialized from
      Provide --clone to copy everything, including ACLs, and not prepend $Organization
      None, realtime, and every-n options for queue GC
      Basic importer
      Allow preservation of ticket IDs durimg import
      PreInflate for various object types
      PostInflate, for columns which depend on our inserted ID
      Make global things not-global
      Pretty graphs while you export
      Syntax fixes
      If we serialize RT::System attributes, we need to resolve it
      Move PostProcess into object serialization instead
      Move logic into RT::Importer rather than sbin/rt-importer
      Add --verbose flag, for object-by-object updates
      Add command-line import option to preserve ticket IDs
      Don't create dependent attributes if we found an equivilant object in the current DB
      Skip Attributes on the system object for now
      Warn on import if --preserve-tickets is passed and there are existing tickets
      Tickets requires _SQLLimit rather than Limit
      Make sure we've loaded Time::HiRes prior to using it
      Warnings removal
      Move PreserveTicketIds into the importer
      Add an option to add the original ticket ID as a CF
      Only warn about existing tickets if we are attempting to preserve ticket ids
      Ensure that we export links to deleted tickets, as well
      Write a summary file, to give importing a sense of what to expect
      Major serializer UI tweaks (progress bars, drop write rate graph)
      Trim some unnecessary decimal places from the GC message
      Add rt-serializer POD, and --help
      Move CF creation into RT::Importer
      Use the more straightforward nstore function, rather than opening the file ourselves
      Split out ObjectCount function, as in RT::DependencyWalker
      Add hooks for a progress callback in the importer, as well
      Progressbars on the importer, based on the summary data
      Move modules around to give a clear place for common methods
      Use RT::Migrate for display code shared between serializer and exporter
      Catch lack of UID during dependency insertion, where we have the referrer
      chmod +x new sbin/ files
      Add a short form of --force
      Add --quiet and --verbose for non-ASCII-art modes
      Make the default message handler be more useful
      Rename ->Dependencies to ->FindDependencies to not conflict with Shredder
      Safeguard if ->Process gets an invalid object
      Catch and warn about missing system users and groups
      Give RT::Topic an Object method to go with its ObjectId  and ObjectType
      Error-proof by checking that there is an Object method to serialize
      Only chase the ParentObj if it exists
      Add classes from RTFM to InitClasses' list
      Tack the necessary package name onto the call to Dumper
      Factor out object creation from the loop
      Move more prep into ->new
      Move unglobal'ing into object variables, as it is state we might need to serialize
      Standardize on one case for internal properties
      Import now just takes a directory -- it's much more straightforard that way
      Store internally current file and position, as well as remaining files
      Serialize internal state when we hit an error
      Make ^C stop and write state when we hit an appropriate point in the loop
      When possible, restore state and resume where we left off
      Fix estimated completion time calculation, objects might be pre-imported
      Add percentage to progress bars
      Don't spew progressbars to non-terminals, and allow --quiet
      POD for rt-importer
      Use Pod::Usage for rt-importer
      Typo fix in rt-serializer POD
      Queue could be 0 on templates and scrips
      OCFs may have an id of 0 if they are applied globally
      Make sure to include deleted rows in estimates
      Store the ticket id, not the effective ticket's id
      Allow links to come before the tickets they reference
      Add two classes that are missing ClassAccessibles
      Articles may have links on them
      Links will not be refs if they're just links to hrefs
      ObjectClasses were not serialized correctly because of a copy/paste typo
      Move reporting of missing UIDs to the text interface
      Store the serializing organization in the data file
      Whether or not to prepend the organization to queues and groups is an import-time decision
      Provide a way to list the contents of a dump
      We care about preserving more than just ticket IDs
      Always serialize all ids, in case we want to use them on import
      Merge duplicate ids on create
      Merge templates by name and/or queue
      Serialize the principalid, so we can authentically recreate it at import time if Overwrite
      Skipping CF -> OCF only makes sense if you're trying to serialize one queue, like early tests did
      Ensure that we serialize old OCFVs, for completeness
      Allow --no-resume, which blows away the partial-import file
      We changed the option name in 8781198
      Remove debug output
      Ensure that we serialize the _decoded_ blob
      Ensure that global ACLs are serialized
      Instead of overwriting a database with initialdata, insert into an empty db
      Simple error-checking on the provided directory for importing
      Ensure that we are email-unique, as well
      When cloning, don't skip duplicate scripaction/conditions on import
      Merge branch '4.2/serializer-importer-cloning' into 4.2/serializer-importer
      The column is called EmailAddress
      Serializer: Refactor file open/close into separate functions
      Importer: Move directory to being a property of the importer, not an arg
      Importer: Do statefile maintenance in the importer itself
      Serializer: Use the same metadata in-stream as in the separate file
      Importer: Unify metadata loading path, including from "rt-serialized"
      Importer: Use ->Metadata abstraction for setting progressbar
      Serializer: Only write stream metadata once
      Serializer/Importer: Export and respect clone property in stream metadata
      Serializer: Clean up gnuplot files and empty logs
      Serializer/Importer: Factor out operations which deal with files
      Serializer/Importer: Move Content and LargeContent into RT::Record
      Serializer/Importer: Simplify clone logic by skipping class methods
      Serializer: Cloning now requires that Principals are serialized
      Serializer/Importer: Allow incremental serialization
      Serializer: Fix NextPage when not cloning
      Prefix the directory to the list of files from the metadata when loading
      Serializer/Importer: Ensure that the total does not exceed the max
      Revert to 3.2-era stylings
      Protect classes with existence checks
      Serializer: Older _ClassAccessible don't include id on Tickets
      Serializer: Perform upgrades on incremental export
      Serializer: Work around bug in sort() in older versions of perl
      Serializer: Keep compat with RT < 3.6 by not assuming RT->can("System") nor RT::System->can("FirstAttribute")
      Serializer: Move record removals before insert/update
      Merge branch '4.2/serializer-importer-incremental' into 4.2/serializer-importer
      Policy fixes
      Merge branch 'master' into 4.2/serializer-importer
      Add incremental upgrade steps for new upgrades
      Merge branch 'master' into 4.2/serializer-importer
      Add incremental upgrade steps for new upgrades
      Merge branch 'master' into 4.2/serializer-importer
      Remove a hard tab that snuck in
      Add recent upgrade steps
      Detection of gpg/openssl should not auto-enable the RT configuration
      Provide better errors during SMIME Probe
      Use the same logic in RT::Crypt::GPG to allow selection of gpg binaries
      Cache the full path to the openssl/gpg binary
      Clear our environment (including PATH) after gpg/openssl have been found
      Allow use of PATH to determine which openssl binary to use
      RT::Test::find_executable is merely File::Which; use it
      Switch the @ACL upgrade (which uses LoadSystemInternalGroup) to a sub
      Ensure that Scrip limits do not assume the current schema
      Enumerate the relevant columns from the schema at the time
      Side-step the ObjectScrips code now found in RT::Scrip::Delete
      Merge branch '4.2/upgrade-from-3.8-errors' into 4.2-trunk
      Old searches over Scrips should not pull in the new Disabled column
      Using RT::Scrip->Create with an in-progress upgrade fails
      Transform flat start/end events into an upgrade hierarchy
      Store the actual return value data structure, to allow for reconstruction
      Make the "after" stage merely metadata; ther is no need to repeat the data
      Store schema and acl events as well
      Recursively walk through the parsed history to display it
      Reconstruct hierarchy on pre-GUID history records
      Ensure that the Attributes table is big enough to contain the upgrades
      Note the upgrades that did not complete
      Re-run license tagger to update years, etc
      Extract PO updates

Jim Brandt (4):
      Hide upgrade history from global attributes on config page
      Add ids to recorded upgrade entries to track success/failure
      Add before entry to insert actions
      Merge branch '4.2/fixup-version-history' into 4.2-trunk

Kevin Falcone (11):
      Document --incremental
      Add a quick set of steps for doing an incremental upgrade
      Clean out any lingering joins set in _Init
      Remind people to use rt-validator
      Merge branch '4.2/smime-gpg-defaults' into 4.2-trunk
      Bump up GnuPG's warning about the homedir
      Handle more formats of the Forward template.
      $queue->Lifecycle is now the DB column, which is undef on upgrade
      Merge in my change to a regex from 474aaf6c11
      Merge branch '4.2/serializer-importer' into 4.2-trunk
      Merge branch '4.2/upgrade-from-3.6-errors' into 4.2-trunk

Thomas Sibley (25):
      Autoconfiscate rt-serializer and rt-importer
      Add a convenient --clone flag when serializing
      Document the behaviour of --gc 0 and --gc -1
      Suggest the command to use for setting up the receiving RT database on import
      Importer: Cloning and saving an OriginalID are incompatible
      Importer: Slightly more reuse, slightly less abstraction busting inside
      Importer: Optionally print out serialized object dumps when listing
      Serializer: Collections containing only disabled objects were GC'd early
      Importer: Preserve global CFs when cloning instead of applying individually
      Serializer: Consistently and conveniently push collections onto the stack
      Serializer: Optionally skip translating object references to UIDs
      Serializer: Cloning without dependency walking or ID translation
      Importer: Avoid a postponed update on Principals when cloning
      Serializer: Autoscale only the endpoints of the axes
      Serializer/Importer: Render progress one last time when finished
      Serializer/Importer: Log warnings to a file, with stack traces for errors
      Serializer/Importer: Clone is a flag which is not negatable
      Importer: Report when we come across objects missing from the source database
      Serializer: Support changing the page size for iterating collections
      Cloning: Select pages of records via id > N rather than OFFSET
      Serializer: Error handling for gnuplot
      Importer: Support resuming a clone when a saved state file exists
      Importer: Catch and rethrow errors from Create with more info
      Importer: Optional ways to handle record creation errors
      Importer: warn instead of printing to stderr on die

-----------------------------------------------------------------------


More information about the Rt-commit mailing list