Fix for AttributeError in _compute_group_range for linked plots

[csachs]

After an update, previously working code produced AttributeError: 'tuple' object has no attribute 'get' exceptions.

Code to reproduce:

import numpy as np
import holoviews as hv
from holoviews.selection import link_selections
import panel as pn

hv.extension('bokeh')

data = np.random.random((1000, 100))
scatter = hv.Scatter(data)
link = link_selections.instance()
example = link(scatter + scatter)

pn.serve(example)

In the plot, select a region in one of the linked plots, then the following exception will occur. It will also directly occur, if the plots are datashaded:

Traceback (most recent call last):
  File "site-packages/tornado/ioloop.py", line 741, in _run_callback
    ret = callback()
  File "site-packages/tornado/ioloop.py", line 765, in _discard_future_result
    future.result()
  File "site-packages/bokeh/server/session.py", line 71, in _needs_document_lock_wrapper
    result = await result
  File "site-packages/tornado/gen.py", line 216, in wrapper
    result = ctx_run(func, *args, **kwargs)
  File "holoviews/holoviews/plotting/bokeh/callbacks.py", line 466, in process_on_event
    self.on_msg(msg)
  File "holoviews/holoviews/plotting/bokeh/callbacks.py", line 163, in on_msg
    raise e
  File "holoviews/holoviews/plotting/bokeh/callbacks.py", line 155, in on_msg
    Stream.trigger(streams)
  File "holoviews/holoviews/streams.py", line 171, in trigger
    subscriber(**dict(union))
  File "holoviews/holoviews/streams.py", line 834, in perform_update
    self.event()
  File "holoviews/holoviews/streams.py", line 415, in event
    self.trigger([self])
  File "holoviews/holoviews/streams.py", line 171, in trigger
    subscriber(**dict(union))
  File "holoviews/holoviews/streams.py", line 919, in perform_update
    self.event()
  File "holoviews/holoviews/streams.py", line 415, in event
    self.trigger([self])
  File "holoviews/holoviews/streams.py", line 171, in trigger
    subscriber(**dict(union))
  File "holoviews/holoviews/streams.py", line 834, in perform_update
    self.event()
  File "holoviews/holoviews/streams.py", line 415, in event
    self.trigger([self])
  File "holoviews/holoviews/streams.py", line 171, in trigger
    subscriber(**dict(union))
  File "holoviews/holoviews/plotting/plot.py", line 244, in refresh
    raise e
  File "holoviews/holoviews/plotting/plot.py", line 240, in refresh
    self._trigger_refresh(stream_key)
  File "holoviews/holoviews/plotting/plot.py", line 257, in _trigger_refresh
    self.update(key)
  File "holoviews/holoviews/plotting/plot.py", line 981, in update
    item = self.__getitem__(key)
  File "holoviews/holoviews/plotting/plot.py", line 446, in __getitem__
    self.update_frame(frame)
  File "holoviews/holoviews/plotting/bokeh/element.py", line 2388, in update_frame
    ranges = self.compute_ranges(range_obj, key, ranges)
  File "holoviews/holoviews/plotting/plot.py", line 638, in compute_ranges
    self._compute_group_range(group, elements, ranges, framewise,
  File "holoviews/holoviews/plotting/plot.py", line 873, in _compute_group_range
    ids = values.get('id')
AttributeError: 'tuple' object has no attribute 'get'

Even without working myself thru the whole code, the offending line looks slightly misplaced:

holoviews/holoviews/plotting/plot.py

Line 873 in 39738bc

ids = values.get('id')

Accessing values here looks misplaced, which might be None, and an if-statement taking care of the case that values is None (which can currently not reached since the access to .get(...) would error in case values being None.

By moving the access past the check, and letting the if handle the tuple case as well, the problem seems solved and the plots are apparently working as desired, however probably someone with more in-depth knowledge of the DimensionedPlot._compute_group_range-function should check if this properly solves the handling.

[philippjfr]

Thanks, I'll try to get this tested and merged by tomorrow and this will likely necessitate a 1.14.1 release asap.

[github-actions]

This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.