Release notes for 0.8.3
=======================

- fixed a bug in HsParser (line numbers were messed up by character literals)


Release notes for 0.8.2
=======================

- incorporated some patches for Martin Norbäck:
    + support for hierarchical module names
    + support for extended pattern guards
    + HTML/CSS fix

- some small fixes, especiall for universal quantification ("forall ...")


Release notes for 0.8.1
=======================

- some language extensions (foreign import/export, pattern type
  signatures, forall keyword).


Release notes for 0.8.0
=======================

- support for new comment format with end-of-line comments

- scanner and parser new analyze the documentation "text", so only
  a defined HTML subset is accepted as input (required to support
  other backends than HTML in the future).

- standard Hugs installations cannot run HDoc, see INSTALL for more
  information!

- support for building on Cygwin with GHC or NHC98.


Release notes for 0.7.3a
========================

- integrated Malcolm Wallace's patch to support for NHC98 


Release notes for 0.7.3
=======================

- "--exports" now really traces import/export relationships between modules,
  so re-exported objects should now appear in the right place.

- indexes of datatypes, classes and instances are produced.

- a couple of bug fixes


Release notes for 0.7.2
=======================

- type synonyms are (finally) supported!

- the broken links problem (links to invisible data types, classes, or
  instances) should be gone

- module loading bugs have been fixed


Release notes for 0.7.1
=======================

- HsParser has been updated, so Haskell '98 should be supported
  entirely now.

- bugs with the "--exports" switch habe been fixed.

- setting the search path with "-I PATH" or "-i PATH" is now possible.


Release notes for 0.7.0 
=======================

HDoc 0.7.0 differs a lot from the early snapshot available before the
0.7.0 release. It won't understand the old input format. I made that
change since HDoc's old input format required *a lot* redundancy
concerning type signatures and class/instance declarations. HDoc now
uses (parts from) HsParser, which enables HDoc to take much more
information from the source code than from its special comments.

The drawback of this is that the input files must be parseable by
HsParser; that implies that no special features like "foreign import",
"need_primHugs", or whatever are used (hopefully HsParser wil be able to
parse that some time in the future).

HDoc's internal structure has also changed radically, so the front end
is now much better separated from the HTML generation backend.