#acl PaulHowarth:read,write,admin,revert,delete All:read

=== Thursday 21st April 2011 ===

==== Fedora Project ====

 * Pseudo-merged all active branches of `perl-Module-Signature` together: `master`, `f15` and `f14` are actually merged together, `f13` has changelog differences only, `el6` and `el5` are merged together and have only changelog differences with `master`, and `el4` is the only branch with a non-changelog difference - the absence of buildrequires/requires for `perl(Digest::SHA)`, which isn't available in EPEL-4

==== Local Packages ====

 * Updated `curl` to fix problem with https via http proxy falling back to http (see http://curl.haxx.se/mail/lib-2011-04/0134.html); there's due to be a new upstream release for this issue tomorrow
 * Updated `perl-CPAN-Changes` to 0.17:
  * Eliminate extra whitespace when release data is not defined ([[CPAN:67441|CPAN RT#67441]])
  * Require `version.pm` 0.79, which introduced the `$LAX` regexp ([[CPAN:67613|CPAN RT#67613]])
  * Add the option to sort groups
 * Updated `python-pyasn1` to 0.0.13b:
  * Major codec re-design
  * Documentation significantly improved
  * ASN.1 `Any` type is now supported
  * ASN.1 `Real` type is now supported properly
  * All example ASN.1 modules moved to separate `pyasn1-modules` package
  * Fix to initial sub-OID overflow condition detection in encoder
  * `BitString` initialization value verification improved
  * The `Set`/`Sequence.getNameByPosition()` method implemented
  * Fix to proper behaviour of `PermittedAlphabetConstraint` object
  * Fix to improper Boolean substrate handling at CER/DER decoders
  * Objects of Constructed types now support `__setitem__()`
  * `Set`/`Sequence` objects can now be addressed by their field names (string index) and position (integer index)
  * Typo fix to `ber.SetDecoder` code that prevented guided decoding operation
  * Fix to explicitly tagged items decoding support
  * Fix to `OctetString.prettyPrint()` to better handle non-printable content
  * Fix to `repr()` workings of `Choice` objects
  * Changes towards performance improvement:
   * All `dict.has_key()` and `dict.get()` invocations replaced with modern syntax (this breaks compatibility with Python 2.1 and older)
   * `Tag` and `tagset` caches introduced to decoder
   * Decoder code improved to prevent unnecessary `pyasn1` objects creation
   * Allow disabling components verification when setting components to structured types; this is used by decoder whilst running in guided mode
   * BER decoder for integer values now looks up a small set of pre-computed substrate values to save on decoding
   * A few pre-computed values configured to `ObjectIdentifier` BER encoder
   * `ChoiceDecoder` split-off `SequenceOf` one to save on unnecessary checks
   * Replace slow `hasattr()`/`getattr()` calls with `isinstance()` introspection
   * Track the number of initialized components of `Constructed` types to save on default/optional components initialization
   * Added a shortcut `ObjectIdentifier.asTuple()` to be used instead of `__getitem__()` in hotspots
   * Use `Tag.asTuple()` and pure integers at tag encoder
   * Introduce and use in decoder the `baseTagSet` attribute of the built-in ASN.1 types

----