Skip to content

Latest commit

 

History

History
140 lines (97 loc) · 4.49 KB

CHANGES.rst

File metadata and controls

140 lines (97 loc) · 4.49 KB

0.4 (unreleased)

  • Numpy >= 2.0 is now supported.
  • The minimum versions required by baseband-tasks are now python 3.10 and baseband 4.2.
  • Remove support for the deprecated astropy test runner. This means it is no longer possible to test the installed baseband package with baseband_tasks.test() inside python; instead, one should use pytest --pyargs baseband_tasks from the command line.

New Features

  • All tasks involving FFTs and padding now automatically ensure efficient lengths of the transforms. [#249]

API Changes

  • The integration.Stack class is deprecated, because of the name conflict with combining.Stack. It has been renamed to integration.PulseStack. Only the latter will be available under baseband.tasks.

Bug Fixes

  • Ensure subclasses of HDFPayload can be used directly with arbitrary headers. [#245]

Other Changes and Additions

0.3 (2021-12-12)

New Features

  • Added support for applying and inverting polyphase filter banks. [#86]
  • Improve resampling and add a new ShiftAndResample task that can also apply time shifts, including possible time delay phase rotations. [#225]
  • Add a simpler ShiftSamples task that just shifts channels by integer number of samples. [#226, #235, #239]
  • Add ability for incoherent dedispersion with DisperseSamples and Dedispersamples. [#238]
  • Streams can now carry meta-data in a meta attribute. This includes information on frequency, sideband, and polarization, all of which are stored in meta['__attributes__'] entry (like astropy's Table class). [#233]

API Changes

Bug Fixes

  • For PintToas, parameters other than observatory and frequency are no longer passed on to PINT's toa.TOA class (but still properly used in toa.get_TOAs_list). [#235]

Other Changes and Additions

  • All tasks now can deal with incomplete last frames. For PaddedTaskFrame, the task will get a complete frame, but it will overlap with the one-but-last frame and only the non-overlapping part is used. [#219, #234]

0.2 (2021-03-29)

This release depends on baseband 4.0 as this allows us to assume (and document) the existence of baseband.tasks. Like baseband 4.0, it requires python 3.7, astropy 4.0, and numpy 1.17.

New Features

  • Streams can now be sliced, returning a new stream for a more limited time span and/or sample shape. [#192]
  • Streams can be turned into arrays by calling np.asarray(stream). No sanity check on amount of memory is done. [#194]
  • SetAttribute can now also be used to change samples_per_frame, shape or dtype. [#195]
  • All streams now have useful repr. For tasks, this includes information on the underlying streams. [#198]
  • The Noise generator has been upgraded to the Philox bit generator, which can be reset to a previous count. This means that noise streams no longer need to carry state information for every frame, and that they are now independent of the order in which frames are initially accessed. [#209]

API Changes

  • For consistency with usage everywhere else, the polarization argument for Power and Square now refers to the output polarization. This allows it to be set to whatever the user wants. [#198]

Bug Fixes

  • Time will now be reported correctly also for Stack and Integrate even if their sample rate is not in Hz. Times will be accurate to the nearest sample of the underlying stream. [#197]
  • Fix bug where times just on the border of polycos sometimes were wrongly considered outside of the valid range. [#206]

Other Changes and Additions

  • Tasks can now be defined with a number of complete samples (shape[0]) that is not an integer multiple of samples_per_frame, which can be used to avoid losing ends of streams for tasks that can handle dealing with partial frames. [#188]

0.1.1 (2020-07-19)

Update that includes the DOI and for which the README.txt is clean enough for twine.

0.1 (2020-07-19)

Initial release. Project renamed from original scintillometry, but similar in that the documentation still suggests that tasks be imported from the various modules in baseband_tasks, which is the only way they can be used for baseband prior to 4.0.