Accessing features in a cluster group in point cluster symbology

Question

I would like to get a dynamic stat value on attributes using point cluster symbology, like it is exposed in this topic QGIS 3.10 - Point cluster symbology sum attribute Unfortunately, the answer proposed is not truely satisfactory. In fact making a buffer around each fictive regroupment point does not exactly correspond to the cluster group (even by getting the correct dynamic buffer with @map_scale) because a point can be taken in several buffers while it is affected to one unique cluster group. It can easily be shown with a high density of points.

Since it doesn't look possible directly in QGIS, I've been trying, with a Python plugin, to subclass QgsClusterPointRenderer but the drawGroup method cannot be rewritten, and it is not allowed to subclass its parent QgsPointDistanceRenderer. And I think this is because are maintained in compile C++ to guarantee accelerated rendering, so it's not a good idea to expect to be otherwhise.

As a matter of fact, I think the only way to obtain such possibility would be to have an evolution of QGIS by exposing like @cluster_size and @cluster_color, a new variable @cluster_group as a virtual layer containing all features included in the cluster group, which would permit to make any operation wanted on features directly through the expression variable

If I'm wrong and if a possibility already exists, I would be glad to learn it. On the oher way, is there any chance to see this future functionality integrated on next QGIS distribution?

Answer 1

I agree this would be a worthwhile addition (as would be many others).

I don't speak for the developer group, but based on observations on the development roadmap over recent versions, I suspect this falls within the type of targeted improvement that ends up happening in one of two ways:

1. An enthusiast who is also a competent GIS or at least C++ developer implements it for their own use, and then contributes it (via a pull request). The existing developer group will act as sounding board and gatekeeper to make sure it is implemented well, and can be quite helpful in assisting. But the enthusiast does the grunt work and needs to have the skillset to be largely self-sufficient.

2. An end-user organization hires one or more experienced current core developers "by the hour" (more or less, I'm not commenting on hour-vs-day-vs-flat-fee) to do the development, likely more efficiently given their proficiency with the existing code. You can see many such functionality enhancements at the release changelogs at https://changelog.qgis.org/en/qgis/, marked as "This enhancement sponsored by ..." or similar.

As a first step for this or any such idea, I'd suggest:

a) Submitting a feature request at https://github.com/qgis/QGIS/issues. You may get useful feedback, either enthusiasm or suggestions of workarounds/complements, etc. Or you may hear crickets, which means "sounds possible, but no one has time to volunteer for it".

b) If you want to hire someone, ask at (or approach individual denizens of) https://lists.osgeo.org/mailman/listinfo/qgis-developer

c) If you want to do it yourself, make a QEP at https://github.com/qgis/QGIS-Enhancement-Proposals/issues once you have a proposed way forward, to get feedback and recognize obstacles before you invest too much effort.