Kelp 0.9015 released

17 views
Skip to first unread message

Stefan Geneshky

unread,
Aug 8, 2014, 3:44:33 PM8/8/14
to perl...@googlegroups.com
Greetings everyone.

Kelp version 0.9015 has been released, fixing some bugs and introducing a few new features and improvements.
This version is fully backwards compatible, so you do not have to make any changes to your existing code if you wish to upgrade.

Upgrading to this version is recommended for those who use the Routes::Controller router to rebless the application instance, as it fixes a typo which caused the application to die.

Bug Fixes
  • Fix a typo causing a fatal error in Routes::Controller
  • Fix debug and error keywords in Kelp::Less
  • Remove an explicit return undef, per Perl::Critic
New Features and Improvements
  • Allow for route methods outside of base class. It is now possible to add a route that lives outside of the base class by prefixing it with '+'. This works for both Perl and camelized strings.
  • Add config keyword to Kelp::Less
  • Virtualize the logger instantiation. 
    • Add a _logger function in Module::Logger, which creates the logger instance, making it easier to subclass the Logger module.
    • Add a Logger::Simple module, which dumps everything to STDERR. This module is convenient to use during development with Kelp::Less.
  • Update bin/Kelp
    • bin/Kelp --less <name> will now create a psgi file with the specified name (previously it would always create app.psgi).
    • Existing files will not be overwritten, unless --force specified.
  • Simplify Less application boilerplate
    • Add a Config subclass with simplified configuration that will be used by default by any Kelp::Less application. This configuration will not load any modules or middleware, leaving it up to the user to manually load any modules needed.
    • bin/Kelp --less <name> will now only generate the psgi application, and not any other folders, views, templates etc.
  • Downgrade confess to croak and croak to die. 
    • Kelp::Response used confess and croak excessively. Both changed to die, so that the user can control how much stack trace they want via the 'long_error' attribute or the KELP_LONG_ERROR env variable.
As always, thanks to everyone who contributed with code, POD and bug fixes.

Stefan
Reply all
Reply to author
Forward
0 new messages