Replies: 3 comments 1 reply
-
Hmm, |
Beta Was this translation helpful? Give feedback.
-
I got a few hours to have a look today and ended up with a (for now) three step process:
Step 1+2 is gathered is here: #2940 Separating it was just done to make it clearer what's what. They should be merged in one go. So before merging I'll just change the merge target of #2941 to I've built and run the container successfully on my machine. Everything looks OK from the logs at least and I can open Transmission. |
Beta Was this translation helpful? Give feedback.
-
I have no problems with going back to in repo install rather than custom build if it's preferred to use newer base os to skip custom builds. |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Hey guys 👋 I've been AFK for a while and think it's very fun that the project lives on 🙌 Big thank you to @pkishino for keeping stuff alive 🙏 Life just got too busy for a while...
I'm trying to get back on the horse again, and the first task: get dev/master branches in sync. Resolving merge conflicts and tagging a version of the image before and after the merge.
And in that work I started thinking. There are a few conflicts in the GitHub workflow setup. And... Building the
haugene/transmission-builder
into a separate docker image is really nice to reduce the build time of the main image.But... I was wondering:
I'm leaning towards option 1. The main reason being maintainability. We started the custom build of Transmission because of their re-write into major version 4. We wanted to fix some issues with it being slow and having a memory leak. But we had also been on version 4 because of a distro change (if memory serves me right?) so there was something about backwards compatibility.
I think that the core feature of this image is being a plug'n'play solution that just works. And then being a bit slow on release updates will probably add to that stability - and we sacrifice the "bleeding edge" versions.
Checking our current
latest
image (ie the version on master branch):docker run --rm -it haugene/transmission-openvpn:latest bash -c "transmission-daemon --version" transmission-daemon 4.0.5 (a6fe2a64aa)
Comparing to just doing
apt update && apt install transmission-daemon
in a plain ubuntu container:root@179e21d72c6f:/# transmission-daemon --version transmission-daemon 4.0.5 (a6fe2a64aa)
It's actually the same version. Although it is true that on
dev
branch we have 4.0.6.So the question is: Is it important to release an image with transmission 4.0.6 now? Or should we ditch houndreds of lines of code/config and just roll with 4.0.5 from the ubuntu repository?
If we want to have the newest Transmission. Then at least I propose that we extract all that into a separate repository that produces the base image. We already have all the code/config and github actions setup - but that would make it a cleaner separation of concerns. And it should produce a vanilla ubuntu image, just with transmission-daemon already installed. Or the .deb file readily available.
Thoughts?
I'm tagging @pkishino as the ship captain and @ksurl because it looks like you wrote most of the new github build config and extracted the transmission-builder to a separate dockerfile etc.
Beta Was this translation helpful? Give feedback.
All reactions