Re: Status and future of the GDAL GNM and image correlator

Hi Alex,
> More than ten years ago, two features were added to GDAL during GSoC
> projects: the raster correlator [0] and support for the Geographic
> Network Model (GNM) [1]. Both features seem to have remained
> limited/experimental, with almost no activity except for code
> formatting and compilation fixes.
True
> As far as I know, they also are
> rarely included in default builds by major distributions/packages,
Are you sure about that ? The raster correlator is part of the build and 
cannot be disabled easily, and even if GNM can be disabled, it is on by 
default. Debian or conda-forge builds include it for example
> the correlator API is not exposed via SWIG.
True
>
> I wonder what is the status and future of this code? Are there any
> plans to improve and expose it, or is the project leaning toward
> deprecation and removal to reduce maintenance overhead?

I was also wondering lately. We also don't see bug reports or questions 
related to them, so they might be little used.

And the fact that gnmanalyze & gnmmanage haven't been ported to the new 
CLI will not help build awareness around them.

Those features don't get that much in our way so there's no immediate 
need to consider deprecation

> In the past I developed an experimental QGIS plugin that uses GDAL GNM
> via Python bindings and considered exposing the correlator as well.
In the same plugin ?
> I would love to hear your thoughts on the future of these two features
> before investing further into plugins, and potentially to the GDAL.

It might be a chicken & egg problem: not enough exposed ==> not enough 
used ==> no feedback ==> no change.

But what's true is that the original developers are apparently no longer 
available (I pinged recently for a GNM related pull request and got no 
feedback).

Even

-- 
Very grumpy about LLMs: FOSS is about increasing public capital,
not becoming enslaved to private equity of giga corporations
--
http://www.spatialys.com
My software is free, but my time generally not.