Upgrade to Forky and NC 33.0.x

[reetp]

Thanks for the scripts!

I have updated various files if you are interested :-) Zero guarantee that my code is 100% trustworthy!

Some notes:

Use Forky
Use NC 33.0.x client
Use .env
Add a few tools for building, and in the container
Add language support to handle odd filenames eg Product©.jpg etc
Add localtime support
Update .netrc format
Add some extra config settings
Add full path names to save issues
Add multiple sync paths
Add a 'if running then wait' check

I just use this to build and deploy:

podman build -f Dockerfile --network host -t nextcloud .
podman compose up -d occlient

Issues:

Logs still an issue - set 50Mb but they don't seem to update/rotate.

I had some issues with big directories and the client appearing to exit at this rather than loop:

/usr/bin/su - nextcloudclient -c "nextcloudcmd

It then restarts from the entry point - I presume that is because there were errors rather than warnings so it exits with 1 instead of 0 and exits the loop.

I need to try and trace this to see if it can be avoided.

I've done this as a big blob - you may want to break it down a bit. I'll leave it with you.

Thanks again for the inspiration.

[DyonR]

Thanks for contributing @reetp! It's always great to see other working on or using code that you've made 😄

First, I would advise against using Forky for this image, since Forky is still Debian testing. For this container I prefer to use stable versions, so Trixie would make more sense to me. However, I do recognize that Trixie currently provides nextcloud-desktop-cmd version 3.16.7-1~deb13u1, while Forky provides 33.0.2-1. That is quite a large jump from the current 3.x version to the new 33.x version, skipping over version 4.x also. I would want to test that more thoroughly before switching.

Besides that, there are a few other problems I've noticed, or things I would personally do differently:
In docker-compose.yml, the image changes from dyonr/nextcloudcmd-client to nextcloud. I do not think that is intended, since this project builds a nextcloudcmd-client image, and does not use the official Nextcloud Docker image.
Also, there seems to be a typo, NEXTCLOUD_PROTO=${NEXTCLOUD:-https}, shouldn't that be NEXTCLOUD_PROTO=${NEXTCLOUD_PROTO:-https}?

(docker-entrypoint.sh) Do not replace PATH with (${NEXTCLOUD_FILEPATH}). This breaks what PATH normally should be, namely PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin. That also seems to be why the script now explicitly uses /usr/bin and /usr/sbin paths for commands like ts, useradd, getent, groupadd, etc. I would rather rely on a proper PATH, or use a different variable name for the Nextcloud file paths.

I would also avoid hardcoding en_GB.UTF-8. Since that locale is region-specific, I think C.UTF-8 would be a better default, neutral encoding, for an image.

Adding an alias ll for ls -l is not something I find nice to include in the container, this could better be used as a post argument in a Docker command if the user wishes to have to.

The Dockerfile also runs COPY ini /root/.config/mc/, but ini does not even exist.

Because of the above, I do not think I can merge this as-is. I do appreciate the contribution though!
I will also definitely look into Nextcloud version 33 to see which new features have been added and which ones could be implemented at a later time.