Plot Rendering Issue in Positron IDE with Ultraplot

[gepcel]

First, I'm unsure if this is specifically an Ultraplot issue or if it's related to using Positron IDE with Python. Feel free to disregard if this isn't relevant to others. However, it's frustrating to have to switch IDEs whenever I need to visualize data during analysis.

Issue Description:
When using Ultraplot with Positron IDE (main website, github), plots fail to render in the plot panel. For example:

import ultraplot as pplt
fig, ax = pplt.subplots()
pplt.show()

Error Message:

Error rendering plot to 'Auto' size: Internal error: Method locally_modified_subplot_params() 
is disabled on ultraplot gridspecs. (-32603)

Additional Context:

• There was a period when this issue disappeared, but I'm uncertain whether this was due to package/IDE version changes or configuration adjustments (e.g., matplotlib backends).
• Current versions:
  • Ultraplot: 1.57
  • Positron IDE: 2025.07.0

[cvanelteren]

I can see in our code that we have this block

    # Delete attributes. Don't like having special setters and getters for some
    # settings and not others. Width and height ratios can be updated with update().
    # Also delete obsolete 'subplotpars' and built-in tight layout function.
    tight_layout = _disable_method("tight_layout")  # instead use custom tight layout
    subgridspec = _disable_method("subgridspec")  # instead use variable spaces
    get_width_ratios = _disable_method("get_width_ratios")
    get_height_ratios = _disable_method("get_height_ratios")
    set_width_ratios = _disable_method("set_width_ratios")
    set_height_ratios = _disable_method("set_height_ratios")
    get_subplot_params = _disable_method("get_subplot_params")
    locally_modified_subplot_params = _disable_method("locally_modified_subplot_params")

[cvanelteren]

So the issue is definitely generated on our end but not the question remains why these methods were disabled by Luke 🤔
The docs from matplotlib state:

locally_modified_subplot_params()
[source]
Return a list of the names of the subplot parameters explicitly set in the GridSpec.
This is a subset of the attributes of SubplotParams.

[gepcel]

So the issue is definitely generated on our end but not the question remains why these methods were disabled by Luke 🤔 The docs from matplotlib state:

There might also be some issues with positron. Searching Error rendering plot to 'Auto' size returns at least 10 issues.
Maybe because it needs to handle like ggplot2 of R, plotly, matplotlib, plotnine, etc.

Anyway, I like the way how Spyder handles this. In contrast, Positron's automatic sizing feature currently produces discrepancies between what appears in the plot panel and the final saved image.

Thank you for taking a look at this.

[cvanelteren]

The interaction is definitely with both: since we are removing the function, we raise an error that proton emits. I am just wondering why we are removing the functions listed. I know that our gridspecs are more flat than matplotlib but outright removing them is a bit strange without proper justification.

[cvanelteren]

Just as a call to action, I am a bit cautious of putting these functions back in. However, I can release the function that is in the error as it merely returns available keywords, have you tried this by any chance?

[gepcel]

Just as a call to action, I am a bit cautious of putting these functions back in. However, I can release the function that is in the error as it merely returns available keywords, have you tried this by any chance?

By commenting this line out # locally_modified_subplot_params = _disable_method("locally_modified_subplot_params"), some simple examples work fine.

[cvanelteren]

That confirms my suspicions. I will need to evaluate if there are potential side effects. The gridspec file is something that is still on my radar to carefully evaluate what is going on there. All I can tell is that it is somewhat at the core for the automatic layout that is applied and it assumes a more flat approach. Will need to do more digging after finishing the other two issue I have at the moment.