Axes.get_shared_x_axes() returning immutable Grouper

raamana · October 20, 2023, 11:52pm

it used to be that Axes.get_shared_x_axes() returned a cbook.Grouper() which had a .join() method, but now it is returning an cbook.GrouperView() that is IMMUTABLE. It is breaking my tool (cant paste the link for some reason) and I don’t even understand the rationale for such backward incompatible change.

regardless, I need a workaround that would continue to work with future versions of matplotlib. Much appreciate your help in pointing me to the right methods and properties to achieve the same effect as:

I can’t paste a link so here is the target: line 528 on diffusion.py module in visualqc library of user raamana on github.

thanks very much.

rcomer · October 21, 2023, 9:34am

Sorry about the GitHub linking problem. It seems that Discourse thinks anyone who posts multiple links to the same site might be a spammer, and you have posted multiple GitHub links (including your previous posts). Looks like there is a setting to alleviate this, but it is above my permissions level, so I’m hoping someone else will change it.

I think this is the link you intended to post. I also see you have since added a commit. So is your problem now solved?

github.com

raamana/visualqc/blob/ba7e77abff7e6fc2adf7ebb6a7fe533ade986fa5/visualqc/diffusion.py#L528


      
              ax.set_frame_on(False)
              ax.set_ylim(auto=True)
              ax.set_ylabel(label, color=color)
              ax.set_zorder(self.layer_order_stats)
              ax.set_alpha(cfg.alpha_stats_overlay)
              ax.tick_params(color=color, labelcolor=color)
              ax.spines['left'].set_color(color)
              ax.spines['left'].set_position(('outward', 1))
          
          # sharing the time point axis
          self.stats_axes[0].get_shared_x_axes().join(self.ax_carpet.xaxis,
                                                      self.stats_axes[0].xaxis)
          # self.stats_axes[0].autoscale()
          
          # 3. axes to show slices in foreground when a time point is selected
          matrix_handles = self.fig.subplots(self.num_rows, self.num_cols,
                                             subplot_kw=dict(rasterized=True),
                                             gridspec_kw=dict(wspace=0.01,
                                                              hspace=0.01))
          self.fg_axes = matrix_handles.flatten()