You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
torchserve-kfs (and perhaps also torchserve) docker images for different python versions
Motivation, pitch
All the torchserve-kfs docker images are currently built with python 3.9.
For users who base their custom images on this image and have dependencies that require 3.10+, this means they can't easily just run an additional pip install command and are required to install a new/additional python version into their custom image or just build the image from scratch.
Also python version 3.9 ends nears its end of life in less than a year (EOL is 2025-10), so this is a good opportunity to make this available.
Alternatives
Either let the user install an additional python version in the image or let the user write a custom image from scratch.
A ready base image with base dependencies + python version would be much more convenient for this
Additional context
No response
The text was updated successfully, but these errors were encountered:
I'd be happy to help integrate this into pytorch/serve, since I believe it would be much better in sync with the rest of the repo there (currently the cron job checks out the repo and git resets hard to the release tag).
I think it would be easier on each release to run that matrix and have all images under the same organization/dockerhubrepo
🚀 The feature
torchserve-kfs
(and perhaps alsotorchserve
) docker images for different python versionsMotivation, pitch
All the torchserve-kfs docker images are currently built with python 3.9.
For users who base their custom images on this image and have dependencies that require 3.10+, this means they can't easily just run an additional
pip install
command and are required to install a new/additional python version into their custom image or just build the image from scratch.Also python version 3.9 ends nears its end of life in less than a year (EOL is 2025-10), so this is a good opportunity to make this available.
Alternatives
Either let the user install an additional python version in the image or let the user write a custom image from scratch.
A ready base image with base dependencies + python version would be much more convenient for this
Additional context
No response
The text was updated successfully, but these errors were encountered: