calculate_hand does not execute immediately

[cirrusasf]

Describe the bug
When I fire up calculate_hand without any argument, it takes a while to show up help information:
usage: calculate_hand [-h] [-v] out_raster vector_file
calculate_hand: error: the following arguments are required: out_raster, vector_file

[asjohnston-asf]

How long did it take for the help message to eventually show up? Seconds, minutes, hours?

I believe @jacquelynsmale was also experiencing this issue with the water_map entrypoint last week.

[jhkennedy]

Yes, this is a problem with the asf-tools container right now (I also meant to open an issue). It's order seconds but 10s of them.

I tried removing this to see if it helped and it does not:
https://github.com/ASFHyP3/asf-tools/blob/develop/Dockerfile#L15

[asjohnston-asf]

@cirrusasf was seeing the delay in a local conda environment, suggesting it's not related to the Docker container.

[jhkennedy]

Oh, interesting. I was only seeing it in the docker container the first time, and not locally for me. 🤔

[asjohnston-asf]

See if bringing your local conda environment up to date introduces the issue? Could have been introduced by a dependency version bump.

[jhkennedy]

Some timing notes:

Both of these commands:

calculate_hand --help
water_map --help

takes ~1.65 seconds in my environment which is fully updated

It's something about the imports in water_map.py and/or hand/caculate.py I think as it takes 0.04s to run:

python -c "import asf_tools; print(asf_tools.__version__)"

but the same ~1.65 seconds to run:

python -c "import asf_tools.water_map; print(asf_tools.__version__)"

[jhkennedy]

On the docker side, the times are even worse the first time they are run ( run each right after a fresh docker run)

calculate_hand --help
water_map --help

takes ~31.5 seconds

python -c "import asf_tools; print(asf_tools.__version__)"

takes ~0.07 seconds

python -c "import asf_tools.water_map; print(asf_tools.__version__)"

takes ~31.5 seconds

---

The second time I run the above commands in a running container, the ~31.5 seconds times drop to ~3 seconds

[jhkennedy]

And completely wiping my local conda environment and recreating sees the same thing as the docker container 31s the first time, 3 seconds the second.

[cirrusasf]

I just tested it. it takes ~3 secs.