Permalink
Switch branches/tags
Nothing to show
Commits on Jan 3, 2012
  1. v3.

    committed Jan 3, 2012
  2. [RT#73666] 5.15.5 changes lvalue substr.

    Previously the limits in the LVALUE were clamped to the extent of the
    string at the point it was created. Now they are not, so our tests need
    updating to account for that.
    committed Jan 3, 2012
Commits on Nov 1, 2011
  1. v2.

    committed Nov 1, 2011
  2. .gitignore dist directories.

    committed Nov 1, 2011
  3. Document the cases where perl forces too early.

    I meant to include this before; good job I'm making another release,
    huh?
    committed Nov 1, 2011
  4. 5.10.0's ~~ has different semantics.

    Since I don't care (anyone using ~~ should obviously be using 5.10.1)
    just skip the test.
    committed Nov 1, 2011
  5. Silence a warning.

    committed Nov 1, 2011
Commits on Oct 24, 2011
  1. v1.

    committed Oct 24, 2011
  2. MANIFEST.

    committed Oct 24, 2011
  3. Support 5.6, since it's trivial.

    Whoever it is keeps inventing these new reftypes really needs their head
    examined. It's *not* helpful.
    committed Oct 24, 2011
  4. Admin stuff.

    committed Oct 24, 2011
  5. Duckumentation.

    committed Oct 24, 2011
  6. >= 5.012, not just >.

    committed Oct 24, 2011
  7. Version-specific test fixes.

    committed Oct 24, 2011
  8. Remove debugging output.

    I wish I could leave this there under $ENV{DEBUG}, but overloading.pm
    doesn't exist on older perls.
    committed Oct 24, 2011
  9. Test strings stay deferred.

    Some of these fail due to bugs in perl:
    
        - Tied scalars evaluating to deferred strings don't defer before
          5.14.
        - Assignment of a 3-part interpolation to an existing lexical forces
          stringification. I've reported this as [perl #101640] but got no
          response yet.
    committed Oct 24, 2011
  10. Get rid of a warning.

    committed Oct 24, 2011
  11. OK, ->join *does* need to be a method.

    I was right the first time: it's a constructor, so it needs to be a
    class method. Keep djoin, as a utility sub that always constructs a
    String::Defer.
    committed Oct 24, 2011
Commits on Oct 23, 2011
  1. Test joining.

    committed Oct 23, 2011
  2. More t::Utils.

        - Move the utility classes into t::Utils, since t/join.t will need
          them too.
        - Supply a glob with a FORMAT entry, and test format refs in
          t/ref.t.
        - Add Test::Exports to our export list.
    committed Oct 23, 2011
  3. Rename ->join to &djoin.

    That is, change it from a method to an exported sub. I suspect that what
    I really want here is a multimethod, but Perl 5 doesn't have those.
    
    The exported sub will always create a String::Defer, even if some of the
    strings it joined were from a subclass. Supporting subclasses properly
    would require a good deal of work: probably the joined string would need
    to hold refs to the strings that had been joined, rather than flattening
    their arrays, and then ->force would need to look through those refs.
    committed Oct 23, 2011
Commits on Oct 22, 2011
  1. Tests for different reftypes.

    At some point I should consider allowing objects with ${} and &{}
    overloading, since they're pretending to be scalar refs and coderefs.
    I'm not sure what to do with objects that have both...
    committed Oct 22, 2011
Commits on Oct 20, 2011
  1. Test .=.

    committed Oct 20, 2011
Commits on Oct 19, 2011
  1. Test the other overloads.

    x and x= currently return a plain value: it should be easy to make them
    defer, though maybe not very cheap. It may be worth providing a complete
    set of string ops at some point, with more structure inside the object
    to support them, but that's a lot more work.
    
    ++ and -- appear to numify the object to its refaddr before operating.
    This is so odd I think it must be a bug in perl: it also happens with an
    object with a trivial stringify overload. If this doesn't get fixed (and
    for perls without the fix, if it does) I think it would be better to
    provide explicit overloads for these two which croak.
    
    .= needs tests in defer.t, and possibly also an implementation.
    committed Oct 19, 2011
  2. Tests.

    I really need to learn how to make test methods like Test::More's, that
    can also be exported as functions.
    committed Oct 19, 2011
Commits on Oct 18, 2011
  1. Initial checkin.

    committed Oct 18, 2011