gdal vector clip: allow to not specify any of bbox, geometry or like …

[rouault]

…to adjust the extent to the one of actual features

Fixes #13519

[PostholerCom]

Thanks alot for doing that, Even!

It may be beneficial to update 'gdal vector rasterize' docs, so people are aware.

[rouault]

It may be beneficial to update 'gdal vector rasterize' docs, so people are aware.

I've actually added your use case as one of the examples

[rouault]

I've actually added your use case as one of the examples

cf https://gdal--13520.org.readthedocs.build/en/13520/programs/gdal_vector_rasterize.html#programs/gdal_vector_rasterize-4

[PostholerCom]

Cool!

Also, rasterize is part of pipeline, now? That's a makes things muuuch easier!

Edit: nevermind, I read that wrong. ;)

[dbaston]

After skimming the implementation, I think I understand what's going on here, but when first reading the example pipeline and documentation it was not clear to me what function clip is performing here. I think that's because this PR is giving clip a new meaning.

If IGetExtent does not provide an extent that takes filtering into account, I think rasterize should either determine the extent on its own, or if this is undesirable because it would prevent streaming, the extent should be updated by gdal vector edit --update-extent or similar.

[PostholerCom]

I think rasterize should either determine the extent on its own, or if this is undesirable because it would prevent streaming, the extent should be updated by gdal vector edit --update-extent or similar.

Perhaps, ! filter should update the extent? filter by design restricts the results, maybe that should automatically update extent by results returned.