precise step

[tiemvanderdeure]

An attempt to avoid the worst floating point errors when calculating step.

@rafaqz let me know what you think about this heuristic

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 82.61%. Comparing base (892688d) to head (fd86424). Report is 10 commits behind head on main.

:exclamation: Current head fd86424 differs from pull request most recent head b852bcb

Please upload reports for the commit b852bcb to get more accurate results.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #656      +/-   ##
==========================================
+ Coverage   82.44%   82.61%   +0.17%     
==========================================
  Files          60       60              
  Lines        4364     4470     +106     
==========================================
+ Hits         3598     3693      +95     
- Misses        766      777      +11     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[rafaqz]

I want to merge this but we probably should have a test of the edge case where the rounding actually does something, just to demonstrate that we do this at all

[rafaqz]

Something weird happening in the resample tests to cause a segmentation violation (signal 11).

https://github.com/rafaqz/Rasters.jl/actions/runs/9148702377/job/25152012187?pr=656#step:5:933

That has to be in GDAL or us doing something bad with a GDAL object

[tiemvanderdeure]

Funilly enough I can't reproduce this by just running tests/resample.jl manually, so not sure what is going on?

[rafaqz]

Maybe its only segfaulting on a specific GDAL version. Welcome to the joy of C pointers

[rafaqz]

@tiemvanderdeure if you can rebase this and push again GDAL will probably work now