Egil Möller

Possibly shortened by using a base64 rather than hex encoding: ``` import uuid, base64 msg_id = base64.b64encode(uuid.uuid4().bytes) ``` We can't just use the raw bytes, as e.g. the json container...

This is how I have compared scipy:s cKDTree to Faiss previously: ``` import faiss import scipy.spatial import numpy as np import pandas as pd import datetime d = 64 #...

To be fair faiss has many other indices than L2, and faiss (and scipy?) are aimed at high dimensional data. It would be nice to have the comparison code checked...

Yeah, faiss.IndexFlatL2 can handle any dimension. I think it's uncommon to use faiss for 2d and 3d data, much more common for > 100 dimensions (word embeddings etc). scipy.spatial.cKDTree can...

I might be late to the game here, but have a look at https://github.com/redhog/pint-proj/blob/master/docs/Example%20usage.ipynb which is something I threw together. It supports converting between geopandas point geometries and pint units...

@hgrecco I'd be happy to play around. I guess an MVP would be: * [ ] PintArray / pint dtypes for non geographic columns in GoeDataFrame * [ ] GeoDataFrame(df)...

@hgrecco My initial tests suggest this "just works", i.e., doing stuff like: ``` >>> a = gpd.GeoDataFrame({"x": [1, 2]}, index=[0, 1], geometry=mygeom).astype({"x": "pint[m]"}) >>> a x geometry 0 1 POINT...

Will have a look at the tests there, that seems promising. I guess my main problem is not knowing enough of the Linux kernel boot sequence to figure out what...

I have to admit I'm a bit stuck... Any help in figuring out what goes wrong would be appreciated. From what I have been able to google, priv=3 combined with...

Why not? If they are nontrivially many? If not, they definitely fall under database rights under EU law, leading to the same issue, that they can not be shared freely...