René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Should zero-sized Rectangles be taken into account for autoscaling purposes?","articleBody":"Currently, Rectangles with zero-width and zero-height are ignored for autoscaling purposes, but not other zero-sized patches:\r\n```python\r\nfrom matplotlib import patches, pyplot as plt\r\n\r\nax = plt.figure().add_subplot()\r\nax.add_patch(patches.Rectangle((-5, -5), 0, 0))  # zero-sized at (-5, -5)\r\nax.add_patch(patches.Circle((5, 5), 0))  # zero-radius at (5, 5)\r\nax.relim()\r\nax.autoscale_view()\r\nplt.show()\r\n```\r\nresults in axes autoscaled around (5, 5) (so they include the circle, but not the rectangle).  I would argue that Rectangles should never be ignored, even if they have zero-width and zero-height, both for consistency with other patches, and because such Rectangles are likely not drawn with no reason.\r\n\r\nThis behavior (implemented in `Axes._update_patch_limits`) was introduced in f24d06c to support log-scaled histograms with empty bins, but is not necessary anymore as hist autoscaling was fully reworked in #14581, making this behavior unnecessary (the check in `_update_patch_limits` can be deleted with no bad effects); perhaps it was even already unnecessary before though.\r\n\r\nAlso, note that #2942 already weakened the original patch by going from ignoring Rectangles with zero width OR height to only ignoring Rectangles with zero width and height.","author":{"url":"https://github.com/anntzer","@type":"Person","name":"anntzer"},"datePublished":"2020-12-07T22:49:47.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":4},"url":"https://github.com/19081/matplotlib/issues/19081"}