Language Mapping Issues
=======================

- Support for "fixed" typecode

- Support for WChar and WString (Owen Taylor writes: ORBit doesn't support
  them, but it should be possible)


ORB Issues
==========

- I don't like to create an extra class for the ORB Object, but we need:

- resolve initial references

- load configurations?

- fixup libIIOP so that g_atexit won't produce a crash when module is
  loaded via PHP dl(module) syntax (see cvs ORBit head -> skip_atexit)

- ...


POA Issues
==========

- I don't think we need a server (POA) interface in PHP scripts, do we?


Configuration issues
====================

- make things configurable (what? orbit configuration?)


PHP issues
==========

- fix up pval, zval mess (reference counting, pointers, ...),
  convert_to_<type>...

- variables are actually modified to reflect the needed data types,
  do conversion on copy of variable

- fix error reporting, exception handling (no try...catch in PHP yet)

- the possibility of making the loaded idl "interface repository"
  somehow persistent

- fix shared module version (current IIOP sucks)

- fix config.m4 (autoconf and especially the PHP extensions are somehow
  hard to understand)
  I think it's already much better now.

- fix PHP so that this ugly hack is not needed with "temporary wrapper"
  classes that are used to make class names lowercase when serializing them

BUGS
====

- lots of testing and fixing

- check these thousands of FIXME

- fix these C++ style // in front of the FIXMEs (yeah, I'm lazy)


Misc
====

- use TODO.xml like in gtk+, I've written an xslt *g*

- check licensing

- write a test suite

- in invocation.c the method call, get and set functions are nearly the same - merge?

- are we destorying the php pointers when we use internal_pointer_reset?

- the request/module cleanup needs to be fixed.
