Permalink
Commits on Sep 5, 2012
  1. v1.953

    melo committed Sep 5, 2012
        * Tweak travis.ci setup
  2. Prepare small release to fix travis.ci setup

    melo committed Sep 5, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  3. Tracking down a problem installing D::Z::PB::Author::MELO:

    melo committed Sep 5, 2012
     * --quiet down most modules installation;
     * --installdeps of D::Z::PB::Author::MELO in a separate step.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
Commits on Sep 4, 2012
  1. Make sure we include perl 5.16

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  2. v1.952

    melo committed Sep 4, 2012
        * Added automatic authentication after connection establishment
        * Support Redis 2.6: Aaron Crane <arc@cpan.org>
        * Attempt to fix non-blocking read on Windows
        * Enable travis.ci support on the repository
  3. Fix dist.ini to work with latest @Author::MELO

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  4. Prepare next release, check Changes

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  5. Be more flexible when detecting Windows:

    melo committed Sep 4, 2012
    We need to know if we are using windows to decide which code to use to set
    a socket to non-blocking behaviour in __fh_nonblocking (used by
    __try_read_sock).
    
    We had reports of blocking on Windows (see #20 and #21), and the
    solution given on #21 is to use replace read() with sysread() in
    __try_read_sock(). The fact that this works is a point in its favor,
    but after that call, we do a ungetc() to put back what we've read.
    
    According to Perl documentation, we should not mix sysread (unbuffered
    I/O) with read/ungetc (buffered I/O) so I don't really like the sysread
    solution.
    
    So I'll try this first instead: the logic behind this commit is that for
    some reason, the socket is not in non-blocking mode when it reaches the
    read() call, and it blocks. This should catch more cases of mswin32 usage,
    and it might fix the issue for good. If it does not, I'll quickly relase
    a sysread-based release but that uses sysread only on Windows (I've tested
    sysread on UNIX systems and it would break as expected).
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  6. Move "is Windows32" decision to a constant

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  7. Our test redis-server can have a customizable timeout now:

    melo committed Sep 4, 2012
     * default is 120 seconds, some tests (eg 04-pipelining.t) would fail
       due to a low timeout value;
     * make sure 07-reconnect.t uses the lowest timeout possible: it depends
       on that.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  8. Make sure our test directory is clean before starting the Redis server

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  9. Make sure REDIS_SERVER_PATH works

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  10. Fix typo in Redis::Hash documentation

    melo committed Sep 4, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
Commits on Jun 18, 2012
  1. Merge pull request #23 from arc/support-2-6

    melo committed Jun 18, 2012
    Support Redis 2.6
Commits on Jun 16, 2012
  1. Document 2.6.x support in Changes

    arc committed Jun 16, 2012
  2. Handle 2.6-format INFO output

    arc committed Jun 16, 2012
    Redis 2.6.x changes its INFO to divide the data into sections, with blank
    lines and comments separating them.  Ensure that we don't let such things
    escape into the data returned.
    
    Tested against Redis 2.6.0-rc5.
  3. Make tests pass under Redis 2.6

    arc committed Jun 16, 2012
    In 2.6.x, the INFO command takes an optional argument; so supply two
    arguments to force an error.  In addition, Redis now yields a different
    error message, so accept either one.
    
    Tested against 2.6.0-rc5.
  4. Make more use of Test::Deep to simplify tests

    arc committed Jun 16, 2012
    It was already one of our test_requires, so this has no additional cost for
    users.
Commits on Jun 9, 2012
  1. Document changes for automatic authentication

    melo committed Jun 9, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  2. Tweak documentation for automatic authentication

    melo committed Jun 9, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  3. Take care of bad passwords:

    melo committed Jun 9, 2012
    If you create your Redis instance with a bad password, it will die with a
    stacktrace. We also disable reconnect to prevent loops.
    
    (updates #11)
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  4. Add support for automatic AUTH command on connect:

    melo committed Mar 12, 2012
    If a password attr is given to new, Redis.pm will automatically send the
    AUTH command after every connection.
    
    First part of #11, but behaviour with auto-reconnect and tests still not
    finished.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
Commits on Mar 14, 2012
  1. Pushed default authority for @Author::Melo dzil bundle

    melo committed Mar 14, 2012
    Will be part of 0.003 of Dist::Zilla::PluginBundle::Author::Melo, so we
    don't need this anymore.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  2. Make sure Dist::Zilla::Plugin::Authority knows who is the AUTHORITY

    melo committed Mar 14, 2012
    I need to see if I can push this to @MELO, it makes more sense in there.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  3. Use --notest while installing Dist::Zilla + our dependencies

    melo committed Mar 14, 2012
    It will be much faster this way, and we don't really care testing the
    dependencies...
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  4. Tweak for travis + Dist::Zilla, using mjgardner/svn-simple-hook example

    melo committed Mar 14, 2012
    Used this .travis.yml file as an example:
    
    https://github.com/mjgardner/svn-simple-hook/blob/master/.travis.yml
    
    Hat tip to #19 (comment)
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  5. Remove --noquiet to debug travis-ci + Dist::Zilla powered modules

    melo committed Mar 14, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  6. Merge pull request #19 from michaelklishin/patch-1

    melo committed Mar 14, 2012
    Fix .travis.yml so that it is valid YAML travis can parse
Commits on Mar 13, 2012
  1. Fix .travis.yml so that it is valid YAML travis can parse

    michaelklishin committed Mar 13, 2012
    Please use lint.travis-ci.org (see [this doc guide](http://about.travis-ci.org/docs/user/travis-lint/)).
  2. Add config file for http://travis-ci.org/

    melo committed Mar 13, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  3. v1.951

    melo committed Mar 13, 2012
        * Remove Tie::StdHash from our dependencies list, its part of core
          and it creates problems with some CPAN clients
  4. Disable POD tests for now, the 404 checks fail with the Github links

    melo committed Mar 13, 2012
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
  5. Remove Tie::StdHash from our dependencies, its creating probs

    melo committed Mar 13, 2012
    Some CPAN clients want to force 5.14.2. on you because of this dependency.
    
    Signed-off-by: Pedro Melo <melo@simplicidade.org>
Commits on Mar 12, 2012
  1. v1.950

    melo committed Mar 12, 2012
        * DEPRECATED: the encoding attribute to new() - we will not support
          automatic decoding after 2.000, so please test your code with
          encoding => undef now.
        * Add pipeline support: Aaron Crane arc@github++!
        * Cache AUTOLOAD calls: improves performance a bit, fixes #2
        * Fix: apply reconnect logic to KEYS and INFO
        * Fix: forbid PING and SHUTDOWN in SUBSCRIBE mode
        * Updated docs covering pipelining and multi/exec
        * Updated docs to point users to Github for code and issues