Feat/hyperspy parity 0.3.0#33
Merged
Merged
Conversation
…e size A host (SpyDE's 3-D IPF window) mounts a scatter3d/voxels figure HIDDEN (display:none) and reveals it later. The first draw3d then ran with the panel at zero drawable size (pw/ph == 0 from a zero-size initial layout), so _gpuInitPanel configured the WebGPU swapchain against a zero-size canvas and the draw pass called getCurrentTexture() on it. On a zero-drawable-size surface that is undefined behaviour: some drivers throw (the mid-draw catch latches p._gpu = 'unavailable', which is terminal because the `_gpu === undefined` init guard never re-runs) and others configure a dead 0x0 context that silently draws nothing. Either way the panel stayed on the Canvas2D fallback forever, even after being revealed at a real size. Fix: - draw3d gates the undefined -> pending transition on pw > 0 && ph > 0, so a zero-size panel keeps _gpu === undefined and re-attempts init on a later draw at the real size (the canvas path renders meanwhile; a hidden panel isn't visible anyway). Applies equally to points and voxels (same init). - A ResizeObserver reveal branch fires redrawAll() when the observed container transitions from zero-size to a real size, so a reveal by a pure display toggle (no layout_json / fig_width model change) still re-attempts init. Verified on a real WebGPU adapter (headless Chromium): a hidden zero-size scatter3d/voxels figure activates WebGPU after reveal, while a normally-visible figure still activates on first draw. A fake strict-driver test proves the terminal 'unavailable' latch no longer fires at zero size even without an adapter. New: tests/test_plot3d/test_gpu_hidden_reveal.py.
Add add_text(x, y, s, ...) to the shared _MarkerMixin, wrapping add_texts for the common single-label case (navigation index, scale-bar label). Returns a _TextHandle exposing set_text / set_color / remove so callers can mutate the label after creation. Backs the HyperSpy anyplotlib backend, whose add_text calls plot.add_text(x, y, s) and relies on those handle methods; previously only add_texts (plural) existed, so the backend fell back to a non-rendering sentinel and nav-index / scale-bar labels never appeared. Adds 12 tests covering both Plot1D and Plot2D, the data/axes/display transforms, set_text / set_color / remove, and transform validation. Claude-Session: https://claude.ai/code/session_018MLraeaoQBWFDScg72W41h
Add an independent right-hand y-axis to Plot1D so overlay curves on a different scale (e.g. a temperature trace over a spectrum) render with their own axis instead of being squashed onto the left scale. Python (plot1d/_plot1d.py): - add_right_axis(color) / remove_right_axis() / set_right_ylabel() / set_right_ylim() / get_right_ylim() - add_line(..., axis="left"|"right"); right lines set right_axis=True and feed a separate auto-range (right_data_min/max), excluded from the left data range so the left scale is unaffected - new state: right_axis, right_y_range, right_data_min/max, right_y_units, right_axis_color Renderer (figure_esm.js): - draw1d computes a second linear scale (right_y_range or auto) and a _toRightY mapper; _drawLine takes an optional y-map so axis=="right" overlays use it - draws the right spine, coloured tick column, and rotated units label in a widened right gutter (only when right_axis is on, so existing 1D plots are unchanged) Tests: 13 unit tests for the state API + a plot1d_twinx visual baseline (headless Chromium) verifying the two scales render independently. Claude-Session: https://claude.ai/code/session_018MLraeaoQBWFDScg72W41h
…ollections Follow-up to the 0.3.0 parity work, addressing code-review findings: - Line1D gains x/data read-back and color/linewidth/linestyle/alpha get+set accessors (backed by the extra_lines entry). Previously an overlay/right-axis line handle exposed none of these, so a consumer reading handle.x crashed and setting handle.color silently no-oped. - Line1D.set_data now recomputes the RIGHT range when the line is a right-axis overlay (it was only ever refreshing the left range, which excludes right lines), so a navigating twinx curve no longer renders against a frozen right scale. - remove_right_axis also resets right_axis_color, and documents that it is a full teardown (a pinned range/label does not survive re-add). - figure_esm.js: the right-axis tick loop iterates the ordered range so an inverted set_right_ylim(hi,lo) still draws tick labels (the flip is still honoured by the line mapping); right gutter widened 46→58px so the rotated units label clears the tick numbers. - markers.py: an empty `lines` collection (segments=[]) is now valid and renders nothing instead of raising "segments must be shape (N,2,2)". Tests: Line1D accessor + right-range recompute + empty-lines cases; twinx visual baseline refreshed for the wider gutter. Claude-Session: https://claude.ai/code/session_018MLraeaoQBWFDScg72W41h
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #33 +/- ##
==========================================
+ Coverage 90.00% 90.18% +0.18%
==========================================
Files 39 39
Lines 3801 3902 +101
==========================================
+ Hits 3421 3519 +98
- Misses 380 383 +3 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Follow up to reach parity with matplotlib for use in hyperspy. Mostly adding in things like cycling colors and adding the ability to twin an axis.