Skip to content
Commits on May 5, 2016
  1. @veryrusty
  2. @veryrusty

    reflect changes

    veryrusty committed May 5, 2016
  3. @veryrusty

    Ensure a request exists before checking for a session

    Sessions are lazy, however we instantiate them before rendering a
    template within a request.
    
    When using `template` as a global keyword, we need to check that a
    request exists *AND* a sesison exists (for the request).
    veryrusty committed Apr 30, 2016
  4. @veryrusty
  5. @xsawyerx @veryrusty

    Make template keyword global (GH #1159):

    This makes sure the `template` keyword is available outside the
    request process. The request object (including the parameters and
    variables) will not be available during that `template` call.
    xsawyerx committed with veryrusty Apr 29, 2016
Commits on May 4, 2016
  1. @veryrusty
  2. @veryrusty

    reflect changes

    veryrusty committed May 5, 2016
  3. @xsawyerx @veryrusty

    Use Module::Runtime::require_module to support older perls better:

    Module::Runtime provides `require_module` in order to support older
    versions of Perl (<= 5.10) better. Since we support 5.8 and 5.10,
    we might as well use this.
    xsawyerx committed with veryrusty May 4, 2016
  4. @xsawyerx @veryrusty

    Replace Class::Load with Module::Runtime:

    Module::Runtime is a faster module than Class::Load (even with
    Class::Load::XS available) and is already available by our stack.
    
    If and when App::Cmd removes its dependency on Class::Load, we
    will not require this by our stack at all.
    xsawyerx committed with veryrusty May 2, 2016
Commits on Apr 29, 2016
  1. @xsawyerx

    v0.166001_02

        [ BUG FIXES ]
        * GH #1160: Engines receive correct log callback on build
          (Peter Mottram)
        * GH #1148: Ensure request body parameter parsing is independent of
          Plack version (Russell Jenkins)
    xsawyerx committed Apr 29, 2016
  2. @veryrusty
Commits on Apr 25, 2016
  1. @veryrusty
Commits on Apr 21, 2016
  1. @veryrusty
  2. @veryrusty

    reflect changes

    veryrusty committed Apr 21, 2016
  3. @veryrusty

    Dont use private methods for body param passing

    The Plack 1.0040 TRIAL release breaks our body param passing; internally
    Plack::Request altered the construction of $env->{plack.request.body}.
    
    Calling the original `body_parameters` method from Plack::Request
    (via SUPER) ensures this is always set up correctly, no matter what
    version of Plack we are subclassing.
    veryrusty committed Apr 2, 2016
  4. @veryrusty
  5. @veryrusty

    reflect changes

    veryrusty committed Apr 21, 2016
  6. @SysPete @veryrusty

    template and session engines receive incorect log_cb on build

    sub _build_serializer_engine already had the correct code:
    
        log_cb => sub { $weak_self->logger_engine->log(@_) },
    
    but template and session were being built with:
    
        log_cb => sub { $weak_self->logger->log(@_) },
    SysPete committed with veryrusty Apr 18, 2016
Commits on Apr 19, 2016
  1. @xsawyerx

    v0.166001_01

        [ BUG FIXES ]
        * GH #1102: Handle multiple '..' in file path utilities.
          (Oleg A. Mamontov, Peter Mottram)
        * GH #1114: Fix missing prereqs as reported by CPANTS.
          (Mohammad S Anwar)
        * GH #1128: Shh warning if optional megasplat is not present.
          (David Precious)
        * GH #1139: Fix incorrect Content-Length header added by AutoPage
          handler (Michael Kröll, Russell Jenkins)
        * GH #1144: Change tt tags to span in skel (Jason Lewis)
        * GH #1046: "no_server_tokens" configuration option doesn't work.
          (Sawyer X)
        # GH #1155, #1157: Fix megasplat value splitting when there are empty
          trailing path segments. (Tatsuhiko Miyagawa, Russell Jenkins)
          NOTE: Paths matching a megasplat that end with a '/' will now include
          an empty string as the last value. For the route pattern '/foo/**',
          the path '/foo/bar', the megasplat gives ['bar'], whereas '/foo/bar/'
          now gives ['bar','']. Joining the array of megasplat values will now
          always be the string matched against for the megasplit.
    
        [ DOCUMENTATION ]
        * GH #1119: Improve the deployment documentation. (Andrew Beverley)
        * GH #1123: Document import of utf8 pragma. (Victor Adam)
        * GH #1132: Fix spelling mistakes in POD (Gregor Herrmann)
        * GH #1134: Fix spelling errors detected by codespell (James McCoy)
        * GH #1153: Fix POD rendering error. (Sawyer X)
    
        [ ENHANCEMENTS ]
        * GH #1129: engine.logger.* hooks are called around logging a message.
          (Russell @veryrusty Jenkins)
        * GH #1146: Cleaner display of error context (Vernon Lyon)
        * GH #1085: Add consistent keywords for accessing headers;
          'request_header' for request, 'response_header', 'response_headers'
          and 'push_response_header' for response. (Russell @veryrusty Jenkins)
    xsawyerx committed Apr 19, 2016
Commits on Apr 13, 2016
  1. @veryrusty
  2. @veryrusty

    reflect changes

    veryrusty committed Apr 13, 2016
  3. @veryrusty
  4. @veryrusty
  5. @veryrusty

    Ensure megasplat keeps empty trailing fields

    The string that a megasplat matched is split on '/'. Without setting an
    LIMIT for the split, empty trailing values get stripped. That is;
    for the route '/foo/**', the paths '/foo/bar' and '/foo/bar/'
    were both giving the splat as [ 'bar' ].
    
    Or, if there were further trailing slashes in the path:
    '/foo/bar///a'  gave  ['bar','','','a'] but
    '/foo/bar///'   gave  ['bar'].   How's that for inconsistent.
    
    Instead, this commit adds a (negative) limit to the split on megasplat
    values so that empty trailing values are kept, which allows you to join
    the arrayref of values together and get the original path again. The
    above examples are:
    '/foo/bar/'     giving ['bar','']
    '/foo/bar///a'  giving ['bar','','','a']  #as before
    '/foo/bar///'   giving ['bar','','','']
    
    This is a subtle change in behaviour, which _may_ cause some Dancers
    issues if they are not aware of the change.
    
    Closes #1155.
    veryrusty committed Apr 13, 2016
Commits on Apr 8, 2016
  1. @xsawyerx

    reflect changes all too late

    xsawyerx committed Apr 8, 2016
  2. @xsawyerx

    Merge branch 'fix/gh-1153'

    xsawyerx committed Apr 8, 2016
  3. @xsawyerx
  4. @xsawyerx

    Remove superfluous header

    xsawyerx committed Apr 8, 2016
  5. @xsawyerx

    Fix GH #1153, correct POD

    xsawyerx committed Apr 8, 2016
  6. @veryrusty

    Update changes

    Add more detail about the merge of the feature/request_response_headers branch
    veryrusty committed Apr 8, 2016
  7. @xsawyerx

    Merge branch 'fastcgi-docs'

    xsawyerx committed Apr 8, 2016
  8. @xsawyerx

    reflect changeg

    xsawyerx committed Apr 8, 2016
  9. @xsawyerx

    Add more detail to CGI and FastCGI Apache options

    Andrew Beverley committed with xsawyerx Jan 28, 2016
  10. @xsawyerx

    Group all Apache config options together

    Andrew Beverley committed with xsawyerx Jan 28, 2016
  11. @xsawyerx

    Remove copyright. This appears to be added later on build, causing du…

    …plicate entries
    Andrew Beverley committed with xsawyerx Jan 28, 2016
Something went wrong with that request. Please try again.