This file documents the revision history for Perl extension Catalyst.
-5.7008 XXXX-XX-XX
+5.7008
+ - Added $c->request->query_keywords for getting the keywords
+ (a query string with no parameters).
+ - Add undef warning for uri_for.
+ - Fix bug where a nested component would be setup twice.
+ - Make ensure_class_loaded behave better with malformed class name.
+ - Make _register_plugin use ensure_class_loaded.
+ - Remove 'Argument "??" isn't numeric in sprintf' warning.
+ (Emanuele Zeppieri)
+ - Fixed a bug where Content-Length could be set to 0 if a filehandle
+ object in $c->response->body did not report a size.
+ - Fixed issue where development server running in fork mode did not
+ properly exit after a write error.
+ (http://rt.cpan.org/Ticket/Display.html?id=27135)
+ - Remove warning for captures that are undef.
+ - Fixed $c->read and parse_on_demand mode.
+
+5.7007 2007-03-13 14:18:00
- Many performance improvements by not using URI.pm:
* $c->uri_for (approx. 8x faster)
* $c->engine->prepare_path (approx. 27x faster)
* $c->engine->prepare_query_parameters (approx. 5x faster)
- - If libapreq2 is installed, URIs are decoded using a C function that is
- approx. 12x faster than URI::Escape.
- Updated HTTP::Body dependency to 0.9 which fixes the following issues:
* Handle when IE sometimes sends an extra CRLF after the POST body.
* Empty fields in multipart/form-data POSTs are no longer ignored.
* Uploaded files with the name "0" are no longer ignored.
- - Added $c->request->keywords for getting the keywords (a query string with
- no parameters).
- Sending SIGHUP to the dev server will now cause it to restart.
- Allow "0" for a path in uri_for.
-
-5.7007 2007-03-13 14:18:00
- Performance and stability improvements to the built-in HTTP server.
- - Don't ignore file uploads if form contains a text field with the
- same name. (Carl Franks)
+ - Don't ignore file uploads if form contains a text field with the same name.
+ (Carl Franks)
- Support restart_delay of 0 (for use in the POE engine).
- Skip body processing if we don't have a Content-Length header.
Results in about a 9% performance increase when handling GET/HEAD