There's been a bunch of enhancements and extensions built as part of this project and it's likely that users of pyqtgraph might eventually want these additions in core at some point. Likely this will be a task more conducive to a modernized upstream code base (eg. type annots, py3.10+, linting on core graphics modules) and likely a full adoption of numba (which maybe is true already?).
2D real-time graphics performance
use of numba, numpy array caching, the m4 downsampling algorithm, QPainterPath pre/append for update, and use of QGraphicsObject.setCacheMode() to massively speed up real-time updated QPainterPath implemented charts.
More or less the reverse chronological set of features was:
https://github.com/pikers/piker/pull/302 which added use of the m4 graphics downsampling algo and sets our official benchmark for insanely fast massive data viz 😎 :
[ ] pretty sure the only outstanding is label stacking from added PlotItem axes?
[ ] submit our updated ._plotitemoverlay.py (or where does core want this put?) and example showing use of link_axes: tuple[int] vs. relay_events: list[str]
talk to pyqtgraph core peeps about this work if they even want it...
[ ] reference the MultiplePlotAxes example and include a detailed write-up on the details of ViewBox.linkView() vs. our new event broadcasting (eg. how you can run into issues with double handling of mouseDragEvent)
Further interesting upstream patches that might pertain improving this API include:
There's been a bunch of enhancements and extensions built as part of this project and it's likely that users of
pyqtgraphmight eventually want these additions in core at some point. Likely this will be a task more conducive to a modernized upstream code base (eg. type annots, py3.10+, linting on core graphics modules) and likely a full adoption ofnumba(which maybe is true already?).2D real-time graphics performance
use of
numba,numpyarray caching, them4downsampling algorithm,QPainterPathpre/append for update, and use ofQGraphicsObject.setCacheMode()to massively speed up real-time updatedQPainterPathimplemented charts. More or less the reverse chronological set of features was:m4graphics downsampling algo and sets our official benchmark for insanely fast massive data viz 😎 :AxisItemlru caching of tick values and a newPlotItem.removeAxis()method were part of a series of patches detailed by and factored intopikerin https://github.com/pikers/piker/pull/413 which originally had an upstream submission in https://github.com/pyqtgraph/pyqtgraph/pull/2160UI / UX extensions
Overlayed
PlotItemview boxes and tiled axes originally proposed in https://github.com/pyqtgraph/pyqtgraph/pull/1359 and put to use from our independent implementation originally proposed in https://github.com/pyqtgraph/pyqtgraph/pull/2162, broken out intopikerin https://github.com/pikers/piker/pull/257 and put to use in https://github.com/pikers/piker/pull/259. Finally, a more final version of thePlotItemoverlay system/API landed in #415.The upstream todo list for this is approx:
pyqtgraphfor https://github.com/pyqtgraph/pyqtgraph/pull/1359 but as an alternative to our original approach in https://github.com/pyqtgraph/pyqtgraph/pull/2162PlotItemaxes?._plotitemoverlay.py(or where does core want this put?) and example showing use oflink_axes: tuple[int]vs.relay_events: list[str]pyqtgraphcore peeps about this work if they even want it...MultiplePlotAxesexample and include a detailed write-up on the details ofViewBox.linkView()vs. our new event broadcasting (eg. how you can run into issues with double handling ofmouseDragEvent)Further interesting upstream patches that might pertain improving this API include:
generalized use of
pyqtgraph.LinearRegionItemand index-allignment between multiple data-source-related graphs was added in https://github.com/pikers/piker/pull/395 and refined to a 2-chart helper routine in https://github.com/pikers/piker/pull/401/files#diff-a91cd3e47717faa8e4317892b26b77102bec85a2f00109e0932cf289d5960d62R812Profiling
We broke out the self-contained
pyqtgraph.debug.Profilerinto a modified version from https://github.com/pyqtgraph/pyqtgraph/pull/2281 and added it topikerin https://github.com/pikers/piker/commit/1d4fc6f327cd2339c26b7a3c83a61fcc6e405c55