Skip to content
Commits on Jan 19, 2014
  1. updating date for 1.54

    committed Jan 19, 2014
Commits on Jan 18, 2014
  1. updating to next version 1.54

    committed Jan 18, 2014
  2. updating date for 1.53

    committed Jan 18, 2014
  3. 1.53

    committed Jan 18, 2014
Commits on Oct 9, 2013
  1. updating to next version 1.53

    committed Oct 9, 2013
  2. updating date for 1.52

    committed Oct 9, 2013
  3. 1.52

    committed Oct 9, 2013
  4. Merge pull request #5 from avar/html-mason-params-correct-docs-for-co…

    …mpiler_class-default
    
    HTML::Mason::Params: Correct lies about complier_class default
    committed Oct 9, 2013
  5. Merge pull request #4 from avar/implement-use_warnings-switch

    Implement use warnings switch
    committed Oct 9, 2013
  6. Merge pull request #3 from avar/do-not-use-global-warnings-in-tests

    Don't turn on global warnings via shebang parsing in tests
    committed Oct 9, 2013
  7. Merge pull request #1 from rjbs/escape-each-item

    escape each part of substitution, not their concatenation
    committed Oct 8, 2013
  8. Merge pull request #2 from alexmv/precedence

    Fix precedence error; "and" is lower precedence than "return"
    committed Oct 8, 2013
  9. 1.52

    committed Oct 8, 2013
Commits on Oct 3, 2013
  1. @avar

    HTML::Mason::Params: Correct lies about complier_class default

    The default is actually HTML::Mason::Compiler::ToObject as noted a few
    lines above, not HTML::Mason::Compiler. Go copy/paste!
    avar committed Oct 3, 2013
Commits on Oct 2, 2013
  1. @avar

    Implement a use_warnings flag for the Mason compiler

    This allows you to catch runtime warnings in all Mason templates by
    enabling a global use_warnings switch, similar to the use_strict
    switch. Unlike use_strict this is off by default, so it's fully
    backwards compatible.
    
    I have a huge codebase where I'd like to enable warnings by default, and
    not have to turn them on one-by-one for each template, which would
    be *really* painful.
    avar committed Oct 2, 2013
  2. @avar

    Don't turn on global warnings via shebang parsing in tests

    A tests in t/06-compiler.t was broken because it relied on $^W being
    true via the shebang being parsed. Mason actually *doesn't* turn on
    warnings for you by default, and these tests were just implicitly
    relying on perl setting $^W because it would parse the shebang of this
    file.
    
    Do a "use warnings" in t/06-compiler.t and other tests instead of
    turning $^W on implicitly via the shebang.
    
    fix tests that were relying on global warnings not to not do
    so (Mason *won't* warn about runtime warnings like "$x + undef" by
    default), and add new tests to show that the previous behavior only
    kicks in if you add an explicit "use warnings".
    
    Remove the shebangs for all tests except t/12-taint.t, where it's
    actually used to enable taint support.
    
    This is needed because I was going to add a use_warnings option similar
    to use_strict, but couldn't actually test for it because it didn't do
    anything, because warnings were globally enabled!
    avar committed Oct 2, 2013
Commits on Sep 27, 2013
  1. @alexmv

    Fix precedence error; "and" is lower precedence than "return"

    Perl 5.19.4 and above produce a warning to detect this common error; it
    rightly detects that use_dhandlers and use_autohandlers incorrectly
    returned true when they were set to the empty string.  The behavior in
    both cases was _equivalent_ to disabling the feature, because the empty
    filename could never be found to be used.
    
    Fix the precedence using explicit parens, and add explicit tests for the
    return values of use_dhandlers and use_autohandlers.  This fixes
    [rt.cpan.org #87050].
    alexmv committed Sep 27, 2013
Commits on Sep 26, 2013
  1. @rjbs
  2. @rjbs

    escape each part of substitution, not their concatenation

    It's easier to think about what the compiled code will do, rather
    than look at the diff alone.
    
    Given this substitution:
    
      <% x(), y(), z() | escape %>
    
    The current code does this:
    
      $m->print(  escape_func( join( q{}, x(), y(), z() ) ) )
    
    After this patch, it does this:
    
      $m->print(  map {; escape_func($_) } ( x(), y(), z() ) );
    
    These are both just slightly simplified.
    
    This change is important because it allows substitutions and escapes
    to interact based on type.  For example, I wrote an escape that
    looks something like this:
    
      sub escape_html {
        my $input = shift;
        return $input->as_html if blessed $input and $input->isa('HTML::Hunk');
        return escape_entities($input);
      }
    
    Then everything is html-escaped unless it explicitly declaring that
    it is HTML.  It means less type-accounting throughout the whole
    application, because the type can always be inspected.
    rjbs committed Sep 24, 2013
Commits on May 9, 2013
  1. updating to next version 1.52

    committed May 8, 2013
  2. updating date for 1.51

    committed May 8, 2013
  3. 1.51

    committed May 8, 2013
  4. add

    committed May 8, 2013
  5. remove hardcoded version

    committed May 8, 2013
Commits on Jul 12, 2012
  1. 1.51

    committed Jul 11, 2012
  2. updating to next version 1.51

    committed Jul 11, 2012
  3. 1.50

    committed Jul 11, 2012
  4. update

    committed Jul 11, 2012
  5. remoe superfluous sections

    committed Jul 11, 2012
  6. revert tidying

    committed Jul 11, 2012
  7. don't perltidy after all

    committed Jul 11, 2012
  8. fix MANIFEST.SKIP

    committed Jul 11, 2012
Something went wrong with that request. Please try again.