RC version | v6.31.2
---------- | -------
Changelog | https://github.com/MichaIng/DietPi/blob/beta/CHANGELOG.txt
Code changes | https://github.com/MichaIng/DietPi/compare/master...beta
v6.31.0 => v6.31.1 | #3639
v6.31.1 => v6.31.2 | #3648
How to apply | https://github.com/MichaIng/DietPi/blob/dev/BRANCH_SYSTEM.md
Release planned | Sunday 05.07.20
dietpi-software, retry it with this new DietPi version as we added Linux builtin module detection which is e.g. true for new Armbian-based images and probably other Linux 5.X images.@MichaIng anything special to test? We still need to talk about test scenarios / catalogue 馃槈
Yes indeed, although I am not sure if/which scenarios are always important to test 馃. I mean we could go through the most critical things:
dietpi-drive_manager mounting and unmounting an external drive, check /etc/fstab that all entries have been (re)created/preserved correctly.I'd say the above is what really must work, everything else is secondary, e.g. can be tested on demand when related scripts have been touched (more then just the comments), e.g.: https://github.com/MichaIng/DietPi/pull/3621/files#diff-620191171f47b4763a532e984acd2e8b (skip .meta scripts and .conf content)
Often all change at least a bid, so more effective would it be go through the changelog and verify that larger changes/mentionable fixes do of course work but as well do not imply any undiscussed regressions/downsides: https://github.com/MichaIng/DietPi/blob/beta/CHANGELOG.txt
Probably when touching any parts of the code, we should maintain a list of affected software titles, exact menus/selections or script call in general. I mean dietpi-config is huge and it hence does not make sense to test all its menus and options manually, but e.g. with this release the RPi Overscan menu, RPi screen resolution menus and RPi sound card selection have been touched + on x86 AMD display drivers have been added. So one can test these specific menus and related options and have an eye on format, spelling, visual glitches and such, and of course the new options in case.
Having some issues with running G_CONFIG_INJECT 'DEV_GITBRANCH=' 'DEV_GITBRANCH=beta' /DietPi/dietpi.txt
Seems like the file /DietPi/dietpi.txt doesn't exist for me.
Issue seems to be covered here https://github.com/MichaIng/DietPi/pull/3629 by using /boot/dietpi.txt instead :smile: which worked great.
Since the changes by @Joulinar only got merged into dev and you've linked to the master.
Other than that the update went smooth and all seems well for my pihole / pivpn, great stuff!
I shall be retrying to use MineOS as I've spent a while trying to get it working pre-beta but never got great results with it, even without using dietpi's software installer.
Edit: wow I just typed up an entire bug report and then I realized I was using the wrong port for mineos. That hurts.
Ah yes a remain from DietPi-RAMdisk time: https://github.com/MichaIng/DietPi/commit/f9232db99a995e7640d58e9ce7aa0159d749c856
Please use: G_CONFIG_INJECT 'DEV_GITBRANCH=' 'DEV_GITBRANCH=beta' /boot/dietpi.txt
I adjusted the link to dev.
@MichaIng
Running some test now. So most probably I will spam you soon. 馃槈
https://raw.githubusercontent.com/MichaIng/DietPi/beta/PREP_SYSTEM_FOR_DIETPI.sh and selected BETA branch[ INFO ] DietPi-PREP | -----------------------------------------------------------------------------------
[ OK ] DietPi-PREP | Step 2: Downloading and installing DietPi source code
[ INFO ] DietPi-PREP | -----------------------------------------------------------------------------------
[ OK ] DietPi-PREP | Checking URL: https://github.com/MichaIng/DietPi/archive/beta.tar.gz
[ OK ] DietPi-PREP | Downloading DietPi sourcecode
[ OK ] DietPi-PREP | Extracting DietPi sourcecode
[ INFO ] DietPi-PREP | Moving kernel and boot configuration to /boot
[ OK ] DietPi-PREP | mv DietPi-beta/config.txt /boot/
[ OK ] DietPi-PREP | mv DietPi-beta/dietpi.txt /boot/
[ OK ] DietPi-PREP | mv DietPi-beta/README.md /boot/dietpi-README.md
[FAILED] DietPi-PREP | mv DietPi-beta/LICENSE.txt /boot/dietpi-LICENSE.txt
mv: cannot stat 'DietPi-beta/LICENSE.txt': No such file or directory
I guess it should be mv DietPi-beta/LICENSE /boot/dietpi-LICENSE or is it a .txt?
Many thanks! Right, the license file has no file ending indeed: https://github.com/MichaIng/DietPi/commit/1cf3b08dae2078a235d6bf0703b75219b88583ac
dietpi.txt.EDIT:
ok, looks more like Chromium installation and not related the update 6.31
It fails to create the link at the end.

AUTO_SETUP_ACCEPT_LICENSE.AUTO_SETUP_ACCEPT_LICENSE inAUTO_SETUP_ACCEPT_LICENSE=1 within dietpi.txt and rebooted my RPi[INFO] DietPi first run setup: Currently running on another screenIs that the intention of the AUTO_SETUP_ACCEPT_LICENSE feature? I thought it was just to accept the LICENSE and not to start the initial setup. If yes, you would need to set AUTO_SETUP_AUTOMATED=1 same time. Otherwise you will be blocked on the first window asking for user input if you are running headless.
The RPi automatically starts the initial setup without user login. Means, no chance to connect via SSH anymore because of the running initial setup
That is strange and not intended without AUTO_SETUP_AUTOMATED=1. The license prompt happens already in the login script, hence the automated login already happened then. Only place where we touch it is here: https://github.com/MichaIng/DietPi/blob/dev/rootfs/var/lib/dietpi/services/dietpi-firstboot.bash#L163
Does the following contain anything?
ls -Al /etc/systemd/system/[email protected]
6.31.1 (beta)Ah now I see your previous test with Chromium auto install. I guess /etc/systemd/system/[email protected]/dietpi-autostart was still present from the last attempt. We do not remove this during PREP (only /etc/systemd/system/[email protected]/autostart known from other images) but we should. I'll also have a look at this Chromium desktop entry creation. Probably this is expected when no desktop is installed, but then we should not attempt to create the symlink 馃槈.
ok I have a new copy burned to my SD card. Will do an initial run on 6.30 and PREP to beta afterwards. Let's see how it behave.
Ok seems you are right. Looks like some leftover from Chromium installation caused the auto login. On a fresh installation, this did not happen.
Okay any drop-in config with autologin inside its name on any local or serial TTY is removed now: https://github.com/MichaIng/DietPi/commit/836b04f9c89850f5df8f1763ac442b622c9b564d
In theory a console on tty5 ([email protected]) or serial console ttyS2 ([email protected]) or such could be enabled with autologin override, which would lead to firstrun setup (automated or prompt) on a console that might not even attached, just like in your case but in case of a serial console more difficult to resolve and in case of tty > 1 one would need to know to hit F<ID> on keyboard to attach the screen to this TTY.
We cannot check/set for all possible things on a pre-image but for this one we have two cases (previous DietPi image with any autologin autostart option enabled + RPi OS autologin on first boot), so lets cover all possible ones.