David Marteau
David Marteau
@michaelkirk thanks for the review > Outside of a sys-crate, I'm surprised for methods to become defined or not based on something other than a feature flag. The problem here...
Do you know if it possible to detect in the build.rs script at compile time if the feature 'with_geos' is activated by the user ? Or should we panic at...
I see an issue: how do you enable feature in tests ? How can we tests the geos methods if only enabled by a feature ?
@michaelkirk More I think about this, more I do not see very much difference between a function not available because gdal is not of the required version and function not...
> That's a valid point. Maybe it's not worth it. Imho, this could even be confusing: `feature` is used for leveraging optional capabilities, but in this case it would leverage...
> If you link GDAL dynamically, do you still need GEOS support at compile-time? Could you elaborate ?
> I meant that the build and run-time environment can be different, Yes, that defeat any compile time detection, and the same reasoning apply for the GDAL version number. But...
> It seems reasonable to return a Result I'm not very enthusiastic with that option : 1) It makes the api not very friendly and amha it has the wrong...
@lnicola I get your point, But this is still bothering me: > availability of GEOS is a run-time, not compile-time attribute, so it shouldn't reflect in compile feature flags or...
* Removed `#[cfg(have_geos)]` compilation directive for geometry methods * Predicates return `Result`