Re: GDAL 3.0.1
On 9/11/19 10:38 AM, Even Rouault wrote:
> On mercredi 11 septembre 2019 10:00:44 CEST Sebastiaan Couwenberg wrote:
>> On 9/10/19 9:56 PM, Sebastiaan Couwenberg wrote:
>>> There is still no fix for fiona, and the upstream issue was closed
>>> without linking to the issue where work on GDAL 3 support is actually
>>> done. I'm very unhappy about that, so I'm not going to waste any more
>>> time on fiona, it will get an RC bug for autoremoval from testing.
>>
>> fiona (1.8.6-3) contains a patch I hacked up to make it build
>> successfully with GDAL 3.
>
> Bas,
>
> Assuming this is
> https://salsa.debian.org/debian-gis-team/fiona/commit/e9fc223ecdbcf2eb967c07c6d056e9e112ccadb8,
It is.
> I'm not sure this is sufficient for proper runtime behaviour.
Probably not. I only made the build & tests succeed, and for the latter
a few needed to be skipped because I lack the insight for a proper fix.
> Particularly I believe that
> OSRSetAxisMappingStrategy(srs, OAMS_TRADITIONAL_GIS_ORDER) calls shall be added on each
> object created with OSRNewSpatialReference(), assuming we want Fiona to still use the
> "traditional GIS order" for coordinates.
OSRSetAxisMappingStrategy is not exposed yet. I guess it needs to be
added to fiona/ogrext3.pxd, and to add the OSRAxisMappingStrategy type
somehow.
I don't know enough about cython and the GDAL API to do this correctly.
My take was to modify the GDAL 2.2 code in Fiona and remove the OSRFixup
calls that were removed in 2.5 to make it work with GDAL 3. Additional
changes were required for the data handling because pcs.csv no longer
exists. So, yeah, I think we want Fiona to still use the "traditional
GIS order" for coordinates. That what everybody expects it to do I assume.
> So from a quick grep, in:
> ./fiona/_crs.pyx
> ./fiona/_transform.pyx
fiona/_crs3.pyx is actually used for the build with GDAL 3, so we'll
likely need a fiona/_transform3.pyx (with the GDAL 3 specific changes)
as well.
Kind Regards,
Bas
--
GPG Key ID: 4096R/6750F10AE88D4AF1
Fingerprint: 8182 DE41 7056 408D 6146 50D1 6750 F10A E88D 4AF1
Reply to: