Package hls_utils about aifororcas-livesystem HOT 5 CLOSED

I guess for that I'll need to run InferenceSystem locally to test that everything works.

Had to fight with my installed Python versions a bit, I know it was only tested on 3.7 but couldn't stop trying newer:

• I had little hope that Python 3.10 on Win would work and it didn't because numpy can't be built, I have abandoned it.
• Python 3.9.5 in Ubuntu didn't work because pip can't find torchaudio==0.6.0 (maybe we should update it in the future)
• Python 3.8.10 worked after installing gcc and python3-dev (sudo apt install gcc python3-dev, note that my Ubuntu has 3.8 as default but you might need to specify python3.8-dev) and wheel (python -m pip install wheel)

Local inference worked fine after that.

Now to testing, I have just replaced hls_utils.* imports with orca_hls_utils.* imports and installed the package. After trying to run inference again I have discovered that I made a mistake in orca-hls-utils dependencies and added python-ffmpeg instead of ffmpeg-python. Both packages exist and they are incompatible! Funny thing is that I made the same mistake when I worked with ffmpeg from Python over this summer. So I have fixed this dependency in the package and released a new version 0.0.2. Everything seems to work correctly now.

I'll be sending PR removing the code from this repo and adding a new dependency tomorrow.

from aifororcas-livesystem.

Test repo: https://github.com/Molkree/orcasound-hls-utils
Package on Test PyPI: https://test.pypi.org/project/orcasound-hls-utils/

Some questions:

1. License: I've put MIT and I guess it's alright to keep it.
2. Build system: for now I've just copied stuff from Packaging tutorial, so it's setuptools>=42 and wheel.
3. Readme: just put in two sentences, probably need to add more!
4. Metadata (in setup.cfg):
   1. What should be the package name? For now I've made it orcasound-hls-utils (this is what shows up on PyPI). Related to it is the package name (the one you use when importing in the code, orcasound_hls_utils now) and repository name (orcasound-hls-utils now).
   2. How will the versioning work? Right now the version is specified in the setup.cfg
   3. Who is the author? (left author/email with placeholders)
   4. Short description: "Collection of utilities for getting Orcasound HLS data" now.
5. Should we specify any version ranges for dependencies?

I will transfer the repository under Orcasound organization when it's ready.

P.S. Oh, and I forgot to add requirements.txt, will do now 😅 Added dependencies.

@prakruti, am I moving in the right direction?

from aifororcas-livesystem.

Sounds great @Molkree! I was earlier wondering if it would be possible to publish from the livesystem repo itself but I guess this is the right thing to do is to create a separate repo for it. Thanks!

from aifororcas-livesystem.

I've transferred the repo to Orcasound (it's here), shortened the name a bit and published it on PyPI.

from aifororcas-livesystem.

So one of the next steps should be removing the code from this repo and instead using the package. I guess for that I'll need to run InferenceSystem locally to test that everything works. Correct, @prakruti?

from aifororcas-livesystem.