Image$isFilled() always returns FALSE

[jonathon]

i've made the following changes.

https://github.com/jamovi/jamovi/commit ... 83552cf56e

we're not planning on doing a release very soon, but if you tell me the OS you use, i can make a pre-release version available for you.

you'll want to place in your 0000.yaml file:

minApp: 2.7.27

jonathon

[NourEdinDarwish]

Hi Jonathon,

Thank you so much for the incredibly quick turnaround on this! That is exactly what I was hoping for.

I'm using Windows 11. I would really appreciate a pre-release version so I can test the fix and continue development. I will make sure to add `minApp: 2.7.27` to our `0000.yaml`.

Regarding the dummy model concept:

Using a dummy model wouldn't quite solve the core issues for a few reasons:

1. Cumulative rendering time: While a single, standard plot doesn't take much time (maybe a second), the effect is cumulative. Our jamovi module contains multiple sub-analysis plots (e.g., the main forest plot, a subgroup forest plot, and a leave-one-out plot) within a single analysis interface. This is especially true for iterative sub-analyses like leave-one-out where it loops over the data many times. If the user has several of these sub-analysis plots enabled and changes an unrelated option (like a meta-regression setting), they all get re-rendered. That quickly cascades into 7+ seconds of waiting for plots that haven't actually changed. This performance hit was effectively preventing me from adding any more sub-analyses to the module. Because we have so many sub-analyses, even if a dummy model saved some computational time on each plot, running multiple dummy models on every unrelated option change would still create a noticeable delay.
2. UX / False feedback: When a user clicks an unrelated option and sees 5 different forest plots flash and re-render on the screen, it gives them the false impression that those specific results actually changed.
3. Accuracy of dummy data: To accurately predict the width using dummy data, we would still need to extract the exact string length of the longest study label, account for the width of the confidence intervals, check for missing values, etc. Trying to build a "fast" dummy layout that perfectly matches the real layout dimensions is complex and not nearly as robust as just measuring the final output.

So the ability to truly skip the rendering pipeline when `clearWith` hasn't triggered is a huge quality-of-life improvement for both developers and users!

Thanks again for looking into this and implementing the fix so quickly. Let me know when the Windows installer is ready.

Best,
Nour

[jonathon]

ugh ... too much AI slop ... 1. it would be surprising if a dummy model and plot would take a whole second to produce 2. you don't need to display it. 3 ... maybe ... but none of this matters because we're going a different route.

here's the .27

https://www.jamovi.org/downloads/jamovi ... in-x64.exe

jonathon