make tests more robust to new GDAL / GEOS / PROJ #158
Labels
Milestone
Comments
visr
added a commit
that referenced
this issue
Sep 22, 2023
Ref #158, this is probably due to GEOS 3.12
yeesian
pushed a commit
that referenced
this issue
Oct 1, 2023
* avoid overwriting TestEnum from other test file
Fixes these test warnings:
```
WARNING: Method definition (::Type{Main.TestEnum})(Integer) in module Main at Enums.jl:210 overwritten on the same line (check for duplicate calls to `include`).
WARNING: Method definition namemap(Type{Main.TestEnum}) in module Main at Enums.jl:214 overwritten on the same line (check for duplicate calls to `include`).
WARNING: Method definition typemin(Type{Main.TestEnum}) in module Main at Enums.jl:215 overwritten on the same line (check for duplicate calls to `include`).
WARNING: Method definition typemax(Type{Main.TestEnum}) in module Main at Enums.jl:216 overwritten on the same line (check for duplicate calls to `include`).
WARNING: Method definition _enum_hash(Main.TestEnum, UInt64) in module Main at Enums.jl:222 overwritten on the same line (check for duplicate calls to `include`).
WARNING: Method definition instances(Type{Main.TestEnum}) in module Main at Enums.jl:225 overwritten on the same line (check for duplicate calls to `include`).
```
* don't test FlatGeoBuf with NULL geometry
GDAL 3.7.2 says it's not supported with a spatial index:
```
GDALError (CE_Failure, code 1):
ICreateFeature: NULL geometry not supported with spatial index
Stacktrace:
[1] maybe_throw
@ ArchGDAL C:\Users\visser_mn\.julia\dev\GDAL\src\error.jl:42 [inlined]
[2] aftercare
@ ArchGDAL C:\Users\visser_mn\.julia\dev\GDAL\src\error.jl:59 [inlined]
[3] ogr_l_createfeature
@ ArchGDAL C:\Users\visser_mn\.julia\dev\GDAL\src\libgdal.jl:28286 [inlined]
```
* update coordinate order in test
Ref #158, this is probably due to GEOS 3.12
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Right now when updating the minor version of the binaries, the ArchGDAL tests usually fail due to minor differences. See for instance #157.
It would be better to modify the tests to be less sensitive to these changes.
For PROJ, we already test with
isapproxbut probably need to introduce a larger absolute difference (@test π ≈ 3.14 atol=0.01). One millimeter differences are probably not interesting to us anyway, we just want to make sure PROJ is doing the projection.For GEOS, the tests often rely on WKT string comparison. However the node ordering is subject to changes, as well as the geometry ordering in a GeometryCollection. Hence it is probably better to move away from WKT string comparison, and just test if the geometries are
ArchGDAL.equals, like was done in JuliaGeo/LibGEOS.jl#81.The text was updated successfully, but these errors were encountered: