Baseband

==================

Bug fixes
---------

================

- The guppi reader now supports reading breakthrough listen files
(which have a header padded with 00 to a 512 byte boundary).

- Remove support for the deprecated ``astropy`` test runner. This means it is
no longer possible to test the installed baseband package with
``baseband.test()`` inside python; instead, one should use ``pytest --pyargs
baseband`` from the command line. [529]

Bug fixes
---------

- Avoid numpy>=2.0 deprecation warnings being raised in the ``__array__()``
methods of ``Payload`` and ``Frame.__array__()``. [525]

================

- The minimum versions required by baseband are now python 3.10, astropy 5.1,
and numpy 1.24.

Bug Fixes
---------

- Ensure header equality tests work with numpy 2.0. [523]

- Ensure Mark4 header shapes always are a tuple of int (no numpy ints). [523]

- Avoid duplicated warnings about bad framesets. [523]

==================

Bug Fixes
---------

- GUPPI files written by the VEGAS backend, which, oddly, writes
numbers as strings with quotes around them, can now be read. [515]

==================

Bug Fixes
---------

- DADA files such as those produced with Meerkat, in which the header ends
without an `` end of header`` comment line can now be read. [500]

- Modules of tasks are now properly accessible. E.g., with ``baseband-tasks``
available, ``from baseband.tasks.dm import DispersionMeasure`` works. [508]

==================

Bug Fixes
---------

- Ensure that Mark 5B files with extra bytes at the start (i.e., with a
first frame that starts later) can be read. [489]

- Ensure that Mark 5B frames with corrupted headers are properly recognized
and do not lead to exceptions. [490]

- Add some minimal validation for Mark 4 and Mark 5B ``FileReader``
arguments. [490]