Stuck on an issue?

Another error spotted, in 2.:

dst_profile = cog_profiles.get("deflate")

with MemoryFile() as mem_dst:
    # Important, we pass `mem_dst.name` as output dataset path
    cog_translate("my-input-file.tif", mem_dst.name, profile, in_memory=True)

should be

dst_profile = cog_profiles.get("deflate")

with MemoryFile() as mem_dst:
    # Important, we pass `mem_dst.name` as output dataset path
    cog_translate("my-input-file.tif", mem_dst.name, dst_profile, in_memory=True)

i.e. profile -> dst_profile.

However I assume we can write straight to S3 without needing to write a local file and using cog_translate?

@kylebarron sure I will create a PR, doing some work with rio-cog so might find some more changes to include

i.e. profile -> dst_profile.

👍

However I assume we can write straight to S3 without needing to write a local file and using cog_translate?

I’m not sure to understand, you mean trying to use Rasterio to write directly to S3? I never tried and as far as I remember it wasn’t possible 🤷‍♂️

just re-tried, and I don’t think it’s possible (at least with goal 2.4)

import rasterio
from rasterio.shutils import copy

with rasterio.open("file.tif") as c:
    copy(c, "s3://my-bucket/test.tif", **c.profile)

>>> CPLE_OpenFailedError: Attempt to create new tiff file 's3://my-bucket/test.tif' failed: No such file or directory

edits: yes, rasterio won’t support writing a vsi file: https://github.com/mapbox/rasterio/issues/1159#issuecomment-532800520