sanko (Sanko Robinson)

Spell check and document evaluate(...) method

Remove test which requires non-core module and is silly besides

Documentation cleanup

Hook into mIRC with XS-based methods

Tidy

Simple error handling, magic mIRC variable setup, and trigger the PERL_ONLOAD signal

Remove simple perl4mIRC::eval_string( ... ) wrapper

Don't bother with Win32::API tests

Eval'd stuff is now under the mIRC::eval:: namespace instead of perl4mIRC::Eval:: and it comes with a slew of silly stuff

Export functions in a less stupid way

Do not destroy the interpreter when the dll times out but clean out the old eval'd packages

Add the layer and apply it to STDERR and STDOUT (yay!)

Start the signal with a wide char in case the printed line begins with a number (Testing use only. I may remove this in the future.)

Send a signal with the appropriate file handle name to mIRC

PerlIOmIRC_pushed(...) I came up with was only subtly different than PerlIOBase_pushed(...) which works just fine, so...

This PerlIOmIRC_write(...) is a c+p of PerlIOBasic_write(...) and may be used in the future to popup some sort of input box when the user wants to readline() from STDIN

Declaring new layer as PerlIO_mIRC. I think I'll only need handle read/write and layer push/pop

Define basic PerlIOmIRC layer

Tack on standard set of headers and include perliol.h

Moving the file mapping stuff from Perl to C

Before v1.0, I intend to remove the kludgey tied IO on the Perl side in favor of using clean PerlIO layering on the C side

Catches missing Win32::API dependancy and alerts user with MessageBox

Adding Barbie for A::C::British

When updating changelog or doing RCS, back the current version up and create a new file

Don't force Devel::PPPort on no-devs

Sorting requires and build_requires alphabetically

Adding more succinct dist abstract

Be less strict because some systems looooooove stupidly accurate numbers (even when the math doesn't back them up)

Internally, platforms parse and define colors differently. (X11 uses XParseColor(), for example)

These bad tests /should/ fail on win32 (and do on *nix)

If it works for one widget type, odds are it works for the rest (for now)

Switching these didn't really make sense to me but I think I was trying to pad the test count

More appropriate abstract

Adding FLTK::ValueInput

gcc4 on 64 systems hate casting pointers being cast to int

0.06921_004 is the most recent unstable (but working) build of AF

Adding version number

Leave the base module version requirement to Build.PL

Correcting prerequisite modules' version numbers

Ahhh, the bastard child of Windows and Unix rears its fugly head.

...should at least use the correct name.

Always rewrite config

Don't require a base version in use statement (leave it to Build.PL)

Compile using Quartz drawing instead of X11 on Apple OS X

Update for FLTK 0.0531 API changes

Updating downloads page for current stable versions