Fail force-cube explicitly for missing features

[Florian-Katerndahl]

Previously, an vector file (e.g. GPKG) could be supplied with a corresponding layer name but a feature count of 0. This leads to force-cube getting stuck.

As this was discovered when using force-tile-extent, a check for the return value was added there with an error message indicating to the user what command to re-run to get more information during which step force-cube failed

[Florian-Katerndahl]

Does is_integer check for "emptiness"? Otherwise I don't see why that check is necessary as the field is always an integer 🤔
I can't find the defintion of the function when searching through the repo, unfortunately.

[davidfrantz]

Thanks Florian,

I never thought about that.

I slightly changed your commit though:

• better test for non-zero instead of 1. This way even other error codes will be catched
• for the counts, I changed it such that it is ensured that the variable is actually an integer, and that it should not be less than 1 (i.e., also catching negative numbers)
• changed the test to rely on the bash library functions (I will push the integer test in a minute)

Thanks,
David

[davidfrantz]

Does is_integer check for "emptiness"? Otherwise I don't see why that check is necessary as the field is always an integer 🤔 I can't find the defintion of the function when searching through the repo, unfortunately.

As said, I pushed it after the merge, it is now in the bash library file. Bash does not have data types. If GDAL changes some things, it is not guaranteed that N_FEATURES holds an integer, it could also be some string.