Release v0.4.0

First release prepared for the public repository.

Added

• Animations now persist inside the .toe. The operator implements
  TouchDesigner's saveData()/loadData(), so channels and keyframes survive a
  save and reload with no external file and no Python involved. This replaces
  saving state by hand from the component.
• channel[index] = keyframe, a synonym for update_keyframe(index, keyframe).
  Keyframes come out of a channel as detached copies, so editing one through a
  Keyframe object is a read-modify-write; assignment makes that round trip
  discoverable from the subscript a user already reached for.
• A pytest suite covering the Python bindings, run against a compiled extension
  that builds the real binding sources against a fake TouchDesigner context. No
  TouchDesigner install needed, so it runs in CI.
• A Catch2 suite over the persistence codec, covering round-trip fidelity and
  its rejection of truncated, foreign and out-of-range data.
• A TouchDesigner integration harness (run_td_tests.ps1 / .sh) that runs the
  suites inside a real project and gates its exit code on the result, including
  checks that the cooked CHOP output matches what the channels evaluate to. See
  TESTING.md. AnimationViewCHOP is covered there too, across all five view
  modes -- it has no Python API of its own, so integration is the only place it
  can be tested at all.
• RangeEnd is exposed to Python, as a trailing argument on
  Channel.evaluate_range and Channel.evaluate_range_by_rate. It defaults to
  RangeEnd.EXCLUSIVE; pass RangeEnd.INCLUSIVE when samples are points on the
  curve rather than spans of time, as when plotting or building a lookup table.
• LICENSE, NOTICE, README.md and CONTRIBUTING.md.

Fixed

• Breaking: range mode sized its output as end_time * sample_rate, which
  ignored the range start. A node with Range = [10, 70] at 60 fps emitted 4200
  samples for a 60-second span. Both range and auto-range now take their count
  from Animation::num_samples, so the length and the data cannot drift.
  Output lengths change for any node whose range does not start at zero.
• Breaking: the range is now half-open. A span of n sample periods produces
  n samples spaced exactly 1/rate apart, and the range end is no longer
  sampled — 30 seconds at 60 fps is 1800 samples, not 1801. Both operators now
  fill their output with evaluate_range_by_rate() rather than
  evaluate_range(), which spreads a count across a closed interval and so
  only lands on 1/rate spacing for one particular count. A CHOP's samples are
  implicitly one period apart — the format stores no per-sample times — so the
  previous pairing skewed the whole channel whenever the two disagreed. This
  follows the same change in anim.
• AnimationViewCHOP's samples view was one sample short of its range in seconds
  mode, while its samples mode was correct. Both now clamp to at least one
  sample, so an inverted range cannot ask TouchDesigner for a negative count.
• Range mode assigned the start before the end unconditionally. Because the
  setters clamp against the current opposite bound, a node moving from [0,30]
  to [50,70] clamped the new start against the stale end and cooked [30,70]
  for a frame. It now settles in a single cook, matching what the Python
  start_time/end_time setters already did.
• The Output Mode parameter defaulted to fullrange, which is not one of its
  menu entries; it landed on the first entry by fallback.
• The Sample Rate parameter had its slider bounds swapped (minimum 120,
  maximum 30) on both operators.
• Holding a Channel past a remove_channel() unwound a C++ exception through
  the CPython boundary instead of raising. It now raises RuntimeError.
• NaN in the cooked output. TouchDesigner allocates a CHOP's sample buffer but
  does not initialise it, so any sample an operator does not write keeps
  whatever was in that memory -- which appears as NaN, apparently by design, so
  the omission is visible. Several paths wrote nothing at all: AnimationCHOP in
  Input mode with nothing connected, or with a mismatched input, set an error
  and returned; AnimationViewCHOP did the same with no source operator
  selected, which is the state a freshly created node is in. Those paths now
  write zeros, so a node that cannot produce data reports why rather than
  emitting NaN. The normal paths are unchanged and still write each sample
  exactly once. Closes #15.
• A channel with no keyframes crashed AnimationViewCHOP. The segment count was
  computed as size() - 1 on an unsigned type, so an empty channel wrapped to
  SIZE_MAX and undercounted the segment table, which the fill loop then wrote
  past the end of. Creating a channel before keying it is ordinary, so this was
  reachable from the first thing a user does. Closes #16.

Changed

• Updated to anim v0.4.0, which makes Id's constructor private, returns
  references rather than pointers from its Id lookups, and moves sampling to a
  half-open range.
• Breaking: removed Channel.num_samples(rate), following anim, which
  removed the method behind it in 0.4.0. A channel knows only the extent of its
  own keyframes -- an editing concept, not the range a host samples over -- so a
  count taken from it silently answered about the wrong span. Use the operator's
  num_samples, which counts over the configured range, or TouchDesigner's own
  numSamples on the cooked output.
• CI builds now run the test suites, and publish release archives that work on
  unzip: the operators in a Plugins/ folder beside the example project,
  Keyframer.tox, the modules the project loads, and the licence files.

Documentation

• Documented that Channel is a live handle while Keyframe and Point are
  values, in the type docstrings, on each keyframe accessor, and in docs/.
  Setting a property on a keyframe read from a channel changes only the copy;
  this is forced by anim, where a keyframe has no identity of its own and every
  edit has to be re-solved against its neighbours.

Known limitations

• Restoring an animation from a .toe loses anim's cached pre-inheritance
  function and handle mode for each channel's last keyframe. The cache is
  private to the library, so the codec cannot persist it. The effect is only
  visible if a keyframe is appended after a reload: the formerly-last keyframe
  keeps its inherited function rather than reverting to the one it was created
  with.

What's Changed

• Prepare AnimationCHOP for public release (v0.4.0) by @keithlostracco in #17

Full Changelog: v0.3.1...v0.4.0

v0.3.1

keyframer update (#14)

v0.2.1

Merge pull request #9 from Actualize-Interactive/update-output-name

update output name for Release