livepeer / livepeer/go-livepeer

Use dynamic linking for Tensorflow in go-livepeer

Open
#1,982 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

area: transcoding status: backlog type: enhancement
Dominant language
Go
Stars
586
Forks
226
Avg merge
1d 17h
Merged PRs (30d)
19

Description

https://github.com/livepeer/FFmpeg/pull/7 enables building FFmpeg with Tensorflow support without a hard requirement for the library to be present at runtime. We should update our CI to use the new linking mechanism, and have a single linux build again.

Steps needed to accomplish this:

  • Port the changes to latest livepeer branch of our fork, they currently reside in the livepeer-dynlink branch of our FFmpeg work and don't work with v4.4.
  • Update go-livepeer CI build process to have a single linux build that has tensorflow-enabled but doesn't crash on runtime if it's not present on the system
  • Update LPMS install_ffmpeg.sh script and CI to also support this new mechanism.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start with the linked FFmpeg pull request and the latest livepeer branch, then inspect the go-livepeer CI build process and LPMS install_ffmpeg.sh script. Confirm how the TensorFlow-enabled FFmpeg build is configured in CI. Done means one Linux build works without TensorFlow present at runtime and both go-livepeer and LPMS CI and installation support the new mechanism.

Written by the indexing model from the issue text.

Assessment

Tech stack
go, tensorflow
Domain
build-system, ci-cd
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.