Changelog diff is: diff --git a/Changes b/Changes index 6ed82dd..f57496c 100644 --- a/Changes +++ b/Changes @@ -2,6 +2,29 @@ Revision history for Perl extension Plack Take a look at http://github.com/miyagawa/Plack/issues for the planned changes before 1.0 release. +0.9981 Mon Jul 18 17:24:11 PDT 2011 + [BUG FIXES] + - Plack::Request: Added a sanity check to remove newlines from headers to follow + the PSGI specification #224 + - HTTPParser::PP: Fixed warnings #225 + - plackup now prints errors to psgi.errors rather than STDERR + - Fixes issues with undef returned from streaming handler in middleware #231 + - ContentLength: Do not auto-add Content-Length from block devices, pipes and + character files + + [NEW FEATURES] + - HTTPExceptions: Support ->as_psgi method on exceptions (doy) + - FastCGI: Support psgix.harakiri + + [IMPROVEMENTS] + - Lint: Added more checks to validate header values + - StackTrace: Strip caller information since it is not useful anyway + - HTTPExceptions: Added rethrow option (doy) + - Misc. doc fixes on plackup (chromatic) + - binmode STDIN for CGI handler for Win32 #218 + - Remove the test that tests Server specific handling of Transfer-Encoding + - Fixed POD link (audreyt) + 0.9980 Mon Jun 6 20:24:25 PDT 2011 [BUG FIXES] - Fixed a bug where restarting loader doesn't terminate children (#209)
…handles since -s would fail.
Use a plain L<> for older POD processors.
Add hyperlink to web application frameworks before the L<HTTP::Engine> li
This is kind of a ghetto fix - the response_cb filter is "supposed to" receive undef in the end to indicate that it reaches the end, BUT there's currently no guarantee for the callback filter is called again (call until it returns undef). Fundamentally, the way we reuse the same chunk filter to apply to ARRAY, IO objects and writer is prone to errors and we should probably be more explicit.
…-Encoding HTTP/1.1 spec says the opposite - if Transfer-Encoding is present, Content-Length header MUST NOT be sent, and If a message is received with both a Transfer-Encoding and a Content-Length header, the latter MUST be ignored. At least this test should not belong to the Plack's common test suite.
Perl's newline mangling breaks the upload-parsing in HTTP::Body, so STDIN
…eeds to explicitly read the input as raw data for the test to work in a meaningful manner.
…TDIN needs a binmode. Adding one for STDERR as well, just to be on the safe side.
See https://github.com/miyagawa/Plack/pull/224 for details.
replace LWS with a single SP in header values
…EPL etc. and caller is not a useful information anyway. Added a note for DEBUGGING to use REPL to investigate response values.