Atpy

The main change in this version is that Astropy is now required as a
dependency, and PyFITS, vo, and asciitable are no longer required.

The documentation now includes a migration guide for using astropy.table
instead of ATpy.

Improvements
~~~~~~~~~~~~

- The tests have been re-factored and can now run on Travis-CI
(http://www.travis-ci.org) to ensure stability in future.

Bug fixes
~~~~~~~~~

- Fixed a bug which caused ATpy to crash if a VO table contained no data
(thanks to gellule)

- Allow table type to be unicode (thnaks to ChrisBeaumont)

New Features
~~~~~~~~~~~~

- filenames can now be specified as HTTP or FTP URLs, which will cause
files to be downloaded on-the-fly

- tables can now be read from gzip and bzip2 files for any file-based
format

- tables can now be read from any file object for any file-based format

- added support for LaTeX read/write via asciitable

- PARAM values in VO tables are now read/written as keywords

API Changes
~~~~~~~~~~~

- formats for ASCII output are now Python 2.x formatting specifiers
(instead of tuples).

- added support for memmap= option in FITS tables and table sets

Bug fixes
~~~~~~~~~

- filenames can now be passed as any basestring object, not just str

- when reading from VO tables, the column ID is used instead of the name
since the ID is guaranteed to be unique.

- fixed compatibility with PyFITS 3.0.x

- fixed compatibility with recent versions of h5py

Minor fixes

Removed automatic cleaning of VO column names

Fixed a crash that occurred when importing ATpy if vo was not present

Notes
~~~~~

ATpy is now released under an MIT license

New Features
~~~~~~~~~~~~

- added support for querying the IRSA archive and the Virtual Observatory.
See http://atpy.github.com/format_online.html for more details.

- added support for writing arbitrary ASCII tables using the asciitable module

- added support for copying all column metadata in one go when creating a
new column based on an old column:

>>> t.add_column('b', column_header=t.columns[a])

- read attributes from HDF5 tables

- allow tables to be read/written directly to HDF5 groups

- added a simple HTML writer

API/Internal changes
~~~~~~~~~~~~~~~~~~~~

- use an ordered dictionary for keywords to preserve order

Bug fixes
~~~~~~~~~

- improved handling of booleans in FITS tables

- fixed minor bugs in reading in tables from HDF5 files

- ensure that VO table column names are valid python variable names (patch
provided by Marshall Perrin)