equals? on an RGeo feature raises an error if the rhs isn't also a feature about rgeo HOT 5 CLOSED

Confirmed, this is a bug. I'll get a fix out in RGeo 0.3.1 within the next day.

from rgeo.

Just to point out, I don't think my hack is the way it should be done, but just a way to get my specs to make it a bit farther before bailing :). Seeing as how there are a bunch of callers of Feature.cast(), that should probably be the place to solve this. Come up with the appropriate semantics for passing Feature.cast() a non-Feature and then define equals?, intersects?, crosses?, etc. in terms of that.

Or something. You obviously know the code a lot better than I do ;)

from rgeo.

I think the core issue here is that the == method is currently implemented as an alias of equals?, whereas they shouldn't have exactly the same semantics. The latter is one of the OGC relational predicates and should expect to be passed a geometry. (Otherwise, it should probably raise an ArgumentError or a TypeError.) Feature.cast similarly should expect to be passed a geometry and raise an appropriate exception accordingly. However, the former, ==, has further semantics in Ruby dealing with object equality (and it looks like Rails' fixture code is expecting that), so passing other types should still work (and return false).

The fix I checked in last night makes this separation. The == operator checks the type before passing control to equals?.

from rgeo.

I think the core issue here is that the == method is currently implemented as an alias of equals?, whereas they shouldn't have exactly the same semantics.

Makes sense. I don't have any familiarity with the OGC specs, so I don't have any opinion on that one way or the other.

The fix I checked in last night makes this separation.

Great! I'll look forward to a release with it included!

from rgeo.

RGeo 0.3.1 released with this fix.

from rgeo.