Ryan Grout

@fratambot I have no idea what your production environment is like, but here is my environment spec if you want to recreate it and test. [rasterioleak2.json](https://github.com/rasterio/rasterio/files/14452448/rasterioleak2.json) (the json extension is...

@fratambot I recreated your environment and used your latest version of the script and I still cannot replicate your memory usage on my machine. I was hoping it could be...

@fratambot it's interesting that you installed rasterio via pip. Does that also include building and installing the dependencies of rasterio via pip? When you installed rasterio, was that using a...

I'm still unable to reproduce this behavior on my machine. @fratambot Have you worked with objgraph? https://mg.pov.lt/objgraph/ It could give some insight.

This morning I realized, that I should have benchmarked against main branch with the numpy based transforms. Here are the timings from current main branch: ```python In [8]: %timeit riotransform.rowcol(i,...

Regarding the test failures, a side effect of these changes are that strings get cast to floats when placed in numpy arrays. Should I specifically test for strings and raise...

On possible addition might be to define a special method for AffineTransformer. This would sidestep all of the overhead with checking/creating arrays for transforming a single point. ```python def transform_point(self,...

@sgillies if you have a couple of moments, could you please look over this PR? The test failures are explained here (https://github.com/rasterio/rasterio/pull/3013#issuecomment-1908654085).

ping @sgillies @vincentsarago

@sgillies Using the transform is definitely a more precise method. In my use case, I wasn't dealing with rotated rasters and an approximate check was more than good enough. I...