[gdal-dev] GDAL 3.9.0beta1 available for testing

[Even Rouault]

Sean,
> Rasterio's test suite has 4 errors with GDAL 3.9.0beta1.
>
> Metadata output of gdalinfo has changed.

ok, I've run locally rasterio tests and I see gdalinfo now outputs:

Metadata:
   a=1
   b=2
   AREA_OR_POINT=Area

whereas the test expects

Metadata:
   a=1
   AREA_OR_POINT=Area
   b=2

Order of metadata items shouldn't be relied upon. I believe we have 
changed something internally (likely changing from char** to 
CPLStringList that does sorting) that might indeed have affected it. You 
could also potentially use "gdalinfo -json" output to get something 
easier to parse. But don't you have a rasterio API to get GDAL metadata 
that would avoid using gdalinfo at all?

> As soon as there are docker images I'll look more closely at this.
ghcr.io/osgeo/gdal:ubuntu-full-3.9.0beta1 is already available
>
> Writing a mask to a Rasterio RGB.byte.tif dataset no longer creates a 
> RGB.byte.tif.msk file: 
> https://github.com/rasterio/rasterio/actions/runs/8786444142/job/24109342188#step:8:2231.

Yes, this is intended. Due to this change:

GTiff driver:
  * change default value of GDAL_TIFF_INTERNAL_MASK config option to YES

You might run the test under GDAL_TIFF_INTERNAL_MASK=NO if you want to 
test external mask.

>
> The text of an error message which was "tests/data/corrupt.tif, band 
> 1: IReadBlock failed at X offset 1, Y offset 1: TIFFReadEncodedTile() 
> failed." in 3.8 has changed to "corrupt.tif, band 1: IReadBlock failed 
> at X offset 1, Y offset 1: TIFFReadEncodedTile() failed." This is not 
> a big deal, we can adjust Rasterio's test.
Yes, related to 
https://github.com/OSGeo/gdal/commit/cb5161d907e8b4fa5cb68aaf891c1a258d6475b0
>
> Behavior of SRC_METHOD/DST_METHOD=NO_GEOTRANSFORM for warp options 
> seems to have changed: 
> https://github.com/rasterio/rasterio/actions/runs/8786444142/job/24109342188#step:8:2287. 
> I'm not entirely sure what the intent is in Rasterio's test, will dig 
> into that. I just wanted to say that I see a change in GDAL.

ok, this is related to https://github.com/OSGeo/gdal/pull/9336 , which 
indeed wasn't honoring the SRC_METHOD=NO_GEOTRANSFORM and was using 
directly the geotransform from the source dataset.

Fix queued in https://github.com/OSGeo/gdal/pull/9724

(that said, tests reprojecting from 4326 to 3857 while ignoring the 
geotransform are a bit weird)

Even


-- 
http://www.spatialys.com
My software is free, but my time generally not.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20240422/433c64cb/attachment.htm>