Permalink
Commits on Jan 31, 2013
  1. Added sanity check in the OO API as well when wrap() and mount() are …

    …used at the same time
    miyagawa committed Jan 31, 2013
  2. New preferred Builder API that doesn't pass around apps in ->mount an…

    …d ->to_app.
    
    Also added ->wrap shortcut in case urlmap/mount isn't used.
    miyagawa committed Jan 31, 2013
Commits on Jan 29, 2013
  1. doc cleanup

    miyagawa committed Jan 29, 2013
  2. Merge pull request #366 from karenetheridge/topic/accesslogtimed_and_…

    …runtime
    
    Topic/accesslogtimed and runtime
    miyagawa committed Jan 29, 2013
Commits on Jan 19, 2013
  1. Added documentation

    miyagawa committed Jan 19, 2013
Commits on Jan 18, 2013
  1. cleanup POD doc

    miyagawa committed Jan 18, 2013
  2. Cleanup

    We don't add to AUTHORS via small patches. Dont' take it personally :)
    miyagawa committed Jan 18, 2013
  3. Merge pull request #364 from avar/add-ability-to-pass-env-to-static-m…

    …iddleware-v2
    
    Plack::Middleware::Static: add ability to pass $env to callbacks
    miyagawa committed Jan 18, 2013
Commits on Jan 17, 2013
  1. Plack::Middleware::Static: add ability to pass $env to callbacks

    Change the callback interface for static middleware to pass $env to
    the callbacks in addition to setting $_ and passing $_ as the first
    argument.
    
    Some middleware might want to look at other things in $env, or consult
    some object we previously stuck in $env instead of just looking at
    $env->{PATH_INFO}. This adds the ability to easily do that.
    
    Chat about an earlier version of this patch on #plack which instead of
    just adding $env to the arguments passed to the callaback. See
    #363.
    
        17:29:43 < miyagawa> avar: quite honestly i don't like the pattern you often come up with, that changes callback signature with other options
        17:33:20 < avar> miyagawa: How would you add that to the API while preserving compatibility? Add another option with a sub that has a different signature?
        17:36:55 < miyagawa> avar: yes, or in this case pass $env as a second arg
        17:37:08 < miyagawa> OR, you write a completely new middleware as a subclass
        17:37:35 < miyagawa> i don't know what you actually want to achieve
        17:37:39 < miyagawa> "some middleware might" isn't a use case
        17:38:35 < avar> If you think breaking "my $path = pop" is fine that's fine by me. I was going for something that wouldn't ever break existing stuff without explicitly declaring what you wanted.
        17:40:45 < miyagawa> this is not POE nobody would ever use pop
        17:41:08 < miyagawa> but then you can also use other option like you said
        17:41:48 < miyagawa> and again i don't see a use case
        17:42:36 < avar> I'm setting up a Plack::Request object early in another middleware that everything else has access to, also an object in $env you can log against.
        17:42:51 < avar> Not having to set that up again + being able to log with that object you pass around is handy
        17:48:30 < miyagawa> nobody would use $path = shift either because the interface is $_
        17:48:50 < miyagawa> but don't take that as a suggestion to completely change that
        17:49:22 < avar> Why is it passing $_ in addition to setting it in the first place?
        17:50:05 < miyagawa> i guess no app will break if you don't pass it
        17:50:31 < avar> well some might be doing: my ($path) = @_;
        17:50:51 < doy> which still wouldn't break
        17:51:28 < miyagawa> yeah.
        17:51:41 < miyagawa> the documentation says nothing about the argument anyway
        17:51:42 < miyagawa> BUT
        17:51:56 < miyagawa> i think your use case is pretty odd
        17:52:04 < doy> yeah, the documentation explicitly says "The callback should operate on $_ and return a true or false value."
        17:52:11 < miyagawa> and that it has to be handled separately than adding a new argument to "path:"
        17:52:41 < miyagawa> right
        17:53:04 < miyagawa> I think the easiest compromise would be to pass $path, $env to the callback :)
        17:54:41 < avar> Sure, works for me. For what it's worth I don't usually write APIs with odd semantics like this, I was just going for something that didn't break the documented or implicit API
    avar committed Jan 17, 2013
Commits on Jan 14, 2013
  1. fix blead warning

    doy committed Jan 14, 2013
Commits on Jan 10, 2013
  1. Checking in changes prior to tagging of version 1.0015.

    Changelog diff is:
    
    diff --git a/Changes b/Changes
    index 6d943a0..3575293 100644
    --- a/Changes
    +++ b/Changes
    @@ -1,5 +1,9 @@
     Go to http://github.com/plack/Plack/issues for the roadmap and known issues.
    
    +1.0015 Thu Jan 10 15:19:17 PST 2013
    +    [BUG FIXES]
    +        - Fixed Lint complaining about Latin-1 range characters stored internally with
    +          utf8 flag on (Mark Fowler)
             - HTTP::Message::PSGI::res_from_psgi now always returns empty string
               for an empty response body, so streamed responses are consistent with
               non-streamed (ether)
    miyagawa committed Jan 10, 2013
Commits on Jan 2, 2013
  1. Merge pull request #360 from karenetheridge/topic/streamed_empty_resp…

    …onse
    
    Always return content => '' for empty response bodies
    miyagawa committed Jan 2, 2013
Commits on Dec 18, 2012
  1. Lint now allows Latin-1 range characters even they are with UTF-8

    flags since printing it doesn't cause issues with syswrite etc. Fix #358
    miyagawa committed Dec 18, 2012
Commits on Dec 3, 2012
  1. Checking in changes prior to tagging of version 1.0014.

    Changelog diff is:
    
    diff --git a/Changes b/Changes
    index 0ffe3da..5f9f549 100644
    --- a/Changes
    +++ b/Changes
    @@ -1,5 +1,10 @@
     Go to http://github.com/plack/Plack/issues for the roadmap and known issues.
    
    +1.0014 Mon Dec  3 10:27:43 PST 2012
    +    [BUG FIXES]
    +        - Fixed Hash order in tests for perl 5.17 (doy)
    +        - Fixed StackTrace tests to run with Devel::StackTrace
    +
         [IMPROVEMENTS]
             - Plack::Middleware::AccessLog can now log the worker pid and server
               port (ether)
    miyagawa committed Dec 3, 2012
Commits on Nov 24, 2012
  1. fix hash order dependency for blead

    doy committed Nov 24, 2012
Commits on Nov 20, 2012
  1. Merge pull request #351 from karenetheridge/topic/accesslog_pid

    Topic/accesslog pid
    miyagawa committed Nov 20, 2012
  2. ignore pod scratch files

    karenetheridge committed Nov 20, 2012
Commits on Nov 15, 2012
  1. Checking in changes prior to tagging of version 1.0013.

    Changelog diff is:
    
    diff --git a/Changes b/Changes
    index 835b235..1cd2743 100644
    --- a/Changes
    +++ b/Changes
    @@ -1,5 +1,10 @@
     Go to http://github.com/plack/Plack/issues for the roadmap and known issues.
    
    +1.0013  Wed Nov 14 19:46:49 PST 2012
    +    [BUG FIXES]
    +        - Make sure psgi.input is seeked even when the input is buffered (Getty, leedo)
    +        - Delete invalid (empty) CONTENT_LENGTH and CONTENT_TYPE in FCGI (Getty, leedo)
    +
     1.0012  Wed Nov 14 12:00:17 PST 2012
         [IMPROVEMENTS]
             - Make conditional middleware work with initialization without an app (doy)
    miyagawa committed Nov 15, 2012
  2. Added a seek once again, so that the apps in between might read input…

    … without seeking back
    miyagawa committed Nov 15, 2012
  3. Merge pull request #350 from plack/fcgi-nginx-empty-keys

    Delete invalid fastcgi_param passed from nginx
    miyagawa committed Nov 15, 2012
Commits on Nov 14, 2012
  1. Checking in changes prior to tagging of version 1.0012.

    Changelog diff is:
    
    diff --git a/Changes b/Changes
    index fbacef3..835b235 100644
    --- a/Changes
    +++ b/Changes
    @@ -1,5 +1,10 @@
     Go to http://github.com/plack/Plack/issues for the roadmap and known issues.
    
    +1.0012  Wed Nov 14 12:00:17 PST 2012
    +    [IMPROVEMENTS]
    +        - Make conditional middleware work with initialization without an app (doy)
    +        - Added force option to BufferedStreaming
    +
     1.0011  Sun Nov 11 11:05:30 PST 2012
         [BUG FIXES]
             - Fix bad Content-Length that could be caused with mod_perl (avar)
    miyagawa committed Nov 14, 2012
  2. fix path

    miyagawa committed Nov 14, 2012