Install GDAL via a Python wheel #303
Conversation
brianhelba
force-pushed
the
gdal-install
branch
from
98a1da3 to
6125d48
Compare
| GDAL_LIBRARY_PATH = osgeo.GDAL_LIBRARY_PATH | ||
| GEOS_LIBRARY_PATH = osgeo.GEOS_LIBRARY_PATH |
There was a problem hiding this comment.
| [[tool.uv.index]] | ||
| name = "large_image_wheels" | ||
| url = "https://girder.github.io/large_image_wheels/" | ||
| explicit = true | ||
|
|
||
| [tool.uv.sources] | ||
| gdal = { index = "large_image_wheels" } |
There was a problem hiding this comment.
This is documented for Kitwarians here: https://www.notion.so/kitware-da/uv-20ad1d436ac2801e84b8f0d504710031?source=copy_link#209d1d436ac2800f8756ebabbdd8c450
| "opencv-python-headless", | ||
| "tqdm", | ||
| # guano metadata | ||
| "gdal", |
There was a problem hiding this comment.
I wanted to put this under the correct comment mini-block, but I'm not sure where that would be.
It also seems like geopandas below should go with it?
| g++ \ | ||
| libgdal36 \ | ||
| libc6-dev \ | ||
| libpq-dev \ |
There was a problem hiding this comment.
libpq-dev is no longer needed regardless. It supports Psycopg, which is now available purely via the Python wheel.
brianhelba
force-pushed
the
gdal-install
branch
from
6125d48 to
ef13e23
Compare
|
I think this is an improvement, as I'd like to get towards a place where system libraries aren't required, as this makes deployment much easier and more portable. It also aligns with 12 Factor: Dependencies, as we want all of our application dependencies to be defined within our Python lockfile, rather than implicitly depending on system versions that we may not always have full control over. I need help testing this and need help understanding exactly where GDAL is being required from (so we can make sure that the installed wheel satisfies the usage). |
So we used to have a tile server for the spectrograms until we converted to using side-by-side JPEGS. The other case is in the creation of the GRTS Cell Ids. This is a 10x10km grid across the North America that is used by USGS with their NABat project to identify locations of bats. They originally provided this to us as a large CSV with a tons of rows to be ingested. It's quicker to utilize the shapefiles and process them for ingestion (like 2 minutes vs the 15-20 for the CSV they gave us) |
2 participants
No description provided.