Skip to content
Permalink
Branch: handler-options
Commits on Mar 26, 2009
  1. Improved rackup's ability to handle handler-specific options.

    Graham Batty authored and oncloud committed Feb 27, 2009
    Made it so that you can specify optparse arguments in a handler definition and rackup (and hopefuly any other rack-using frontend) can take
    handler-specific options by dynamically rewriting the optparse table when you specify a server. In theory, this should help alleviate the
    situation where both handlers AND frameworks are writing their own tools to start because they each need to have their own custom options.
    Now handlers can take their custom options through a framework (or rackup's) initializer. Prime example would be rails 2.3's script/server, which
    I hope will also adopt this convention.
  2. Added extra nested params error handling and added messages to except…

    rob-at-thewebfellas authored and manveru committed Mar 18, 2009
    …ions to help with debugging
    
    Signed-off-by: Michael Fellinger <m.fellinger@gmail.com>
  3. Simplified options for unregistered handlers

    Graham Batty authored and manveru committed Feb 26, 2009
    Made it so that it only accepts FullyCapsedClassNames for unregistered handlers as per suggestions in #rack irc channel.
    
    Signed-off-by: Michael Fellinger <m.fellinger@gmail.com>
  4. Made it so that the expectation is that Handler.get raises NameError …

    Graham Batty authored and manveru committed Feb 26, 2009
    …on failure instead of nil under the principle of obvious failure.
    
    Signed-off-by: Michael Fellinger <m.fellinger@gmail.com>
  5. Made auto-registration of unknown handlers much smarter. This should …

    Graham Batty authored and manveru committed Feb 26, 2009
    …make it possible for the responsibility for handler definition to be on the people writing servers without making life difficult for the user. ie. 'gem install thin', if thin includes its rack adapter, would make 'rackup -s thin' just work, which is how it should be.
    
    Signed-off-by: Michael Fellinger <m.fellinger@gmail.com>
Commits on Mar 25, 2009
  1. Push version numbers

    leahneukirchen committed Mar 25, 2009
  2. Update copyright years

    leahneukirchen committed Mar 25, 2009
  3. Revert "Specify PATH_INFO encoding"

    leahneukirchen committed Mar 25, 2009
    This reverts commit 7a3d21f.
  4. Build all files needed

    leahneukirchen committed Mar 25, 2009
  5. Refactor Rakefile

    leahneukirchen committed Mar 25, 2009
Commits on Mar 19, 2009
  1. Added Rack::Request#path

    tj authored and leahneukirchen committed Mar 18, 2009
    I would imagine since its used internally there is a chance other devs will
    need it, and IMO its a mise-well situation, your call :)
    
    Signed-off-by: Christian Neukirchen <chneukirchen@gmail.com>
  2. Refactored Rack::Request#media_type

    tj authored and leahneukirchen committed Mar 18, 2009
    No need for subscript, first reads better
    
    Signed-off-by: Christian Neukirchen <chneukirchen@gmail.com>
Commits on Mar 14, 2009
  1. Rack::Deflater streaming

    rtomayko committed Mar 13, 2009
Commits on Mar 13, 2009
  1. Remove files from params when no file is selected [rack#37 state:reso…

    rob-at-thewebfellas authored and josh committed Mar 13, 2009
    …lved]
    
    Signed-off-by: Joshua Peek <josh@joshpeek.com>
Commits on Mar 12, 2009
  1. Handlers use ContentLength and Chunked middleware where needed

    rtomayko committed Mar 6, 2009
    Each Rack handler now automatically wraps the app in one or more
    pieces of middleware based on how the server is implemented. All
    handlers use the Rack::ContentLength middleware and some handlers
    use the Rack::Chunked middleware. Handlers that don't use the
    Chunked middleware either do not require CE for some reason or
    implement CE at the server level.
    
    * The Thin handler uses Chunked and ContentLength middleware. Thin
      has built-in CE support but also allows the app to apply it
      explicitly. Using our middleware for consistency and also because
      I believe the Thin folks want to remove CE support in the future.
    
    * The Mongrel handler uses Chunked and ContentLength middleware.
      Mongrel has built-in CE support but also allows the app to apply
      it explicitly. Using our middleware for consistency across
      handlers.
    
    * The SCGI handler uses Chunked and ContentLength middleware.
    
    * The WEBrick handler uses ContentLength middleware only; WEBrick
      has a really touchy CE implementation that doesn't like it when
      CE is applied by the application.
    
    * The FastCGI handler uses ContentLength only. FastCGI has its own
      mechanism for transferring bodies that do not include an
      explicit Content-Length. The FastCGI server applies the chunked
      encoding if needed.
    
    * The CGI handler uses ContentLength only. The CGI spec forbids the
      use of the Transfer-Encoding header in output from CGI programs.
      The server program is responsible for applying the chunked
      encoding if needed. Closing stdout signals the end of the body
      so it's not really needed anyway.
    
    * The LSWS handler uses ContentLength middleware only. Like CGI, LSWS
      closes the stream after transfer so chunked encoding is never
      required.
Commits on Mar 11, 2009
  1. Force Litespeed input to be a proper IO object [rack#36 state:resolved]

    russ authored and josh committed Mar 11, 2009
    Signed-off-by: Joshua Peek <josh@joshpeek.com>
Commits on Mar 8, 2009
  1. Specify PATH_INFO encoding

    sam-github authored and leahneukirchen committed Mar 8, 2009
    Signed-off-by: Christian Neukirchen <chneukirchen@gmail.com>
  2. WEBrick handler now sets the unescaped PATH_INFO

    thinkerbot authored and leahneukirchen committed Mar 8, 2009
    Signed-off-by: Christian Neukirchen <chneukirchen@gmail.com>
Commits on Mar 6, 2009
  1. Add Rack::ContentType middleware

    candlerb authored and rtomayko committed Mar 5, 2009
    Signed-off-by: Ryan Tomayko <rtomayko@gmail.com>
Commits on Mar 4, 2009
  1. Quick Rack::Builder code simplification

    candlerb authored and rtomayko committed Mar 4, 2009
    Signed-off-by: Ryan Tomayko <rtomayko@gmail.com>
Commits on Feb 27, 2009
  1. Allow realm to be passed to auth handler's initialize method

    candlerb authored and rtomayko committed Feb 26, 2009
    Signed-off-by: Ryan Tomayko <rtomayko@gmail.com>
Commits on Feb 25, 2009
  1. Use / suffix in text and URLs in directory listing

    Yonghui authored and rtomayko committed Feb 25, 2009
    Signed-off-by: Ryan Tomayko <rtomayko@gmail.com>
  2. Make MethodOverride + Auth::Digest play well together

    S. Brent Faulkner authored and rtomayko committed Feb 5, 2009
    Auth::Digest::Request#method now uses any value in
    rack.methodoverride.original_method to override the value in
    REQUEST_METHOD
    
    Signed-off-by: Ryan Tomayko <rtomayko@gmail.com>
Older
You can’t perform that action at this time.