If the minimal-live CD booted on the macbook, then the mac is capable of booting in legacy bios (CSM) mode. There is no uefi bootloader in that iso. And I agree that it should have installed in virtualbox. I have installed previous versions of the minimal-live in previous versions of virtualbox, and the current isos are essentially the same as they were then.

If you choose expert install in the netinstall iso, you get a few more questions, and one of them lets you skip adding a bootloader. That would have avoided clobbering rEFInd.

I booted the first option in the minimal live. Choosing one of the toram or access options shouldn't change the outcome. I'm not sure about the no-probe option.

This one: devuan_beowulf_3.1.1_amd64_minimal-live.iso

In your virtualbox gui, go to System, Motherboard, Chipset. What does it say?

This is not really making sense to me.

This section starts with line 524. It should tell how many efi partitions are present. I don't understand the error message on that.

if [ "$esp_count" -eq 1 ] ; then
        esp_count=$(env LC_ALL=C fdisk -l | awk '/EFI System/ { print $0 }' | wc -l)
fi
if [ "$esp_count" -gt 1 ] ; then
        must_choose_esp="yes"
fi

The error messages about  'Unable to open /dev-sr0 read-write' and 'Can't have a partition outside the disk' make me think that it's trying to install grub to the wrong disk. (i.e. to the CDROM). Sometimes grub and the kernel don't agree on which disk is first. This might be one of those cases.

Lorenzo,

Thanks for the thorough reply. I'm ok with using the scripts in rcS.d - it works fine that way. I might play around with the more advanced configs later.

I was thinking ahead to the future. Sysvinit scripts are disappearing at an alarming rate, and I worry that they won't all land in the new orphan-sysvinit-scripts package.

If you are installing from the minimal-live, there will already be at least one ext2/3/4 filesystem on the virtual disk when you get to the grub-install stage.  What do fdisk and blkid show for the disk? What virtualization software are you using on the mac? Maybe it does something differently from what I expect.

There's no specific iso for macs, but the amd64 desktop-live has both the 32 and 64-bit bootloaders and is know to work on at least one old macbook pro. If you do in fact need the 32-bit bootloader, you will need to install the grub-efi-ia32 package in the live environment before you install the system. You can install it with apt if you have network, or you can install it with

dpkg -i /grub-efi-ia32*.deb

When you install the package, don't let it install the bootloader at that time. The live installer (Refracta Installer) will let you choose the efi partition if there are two or more.

Make sure you check if that's what you need with

cat /sys/firmware/efi/fw_platform_size

before you try the 32-bit bootloader.

I'm not sure that will be enough. You still need a way to make the machine boot from the usb drive. Google says hold down the option key at boot and then select the usb drive.

A netinstall from a 3.0 iso would give you 3.1 or actually a little newer than that since some packages have been upgraded since last week. Volume settings on my up-to-date mate don't get lost. I do keep losing the hardware setting. Analog Stereo Output keeps reverting to Analog Stereo Duplex.

I tried running 'alsactl store' before rebooting and 'alsactl restore' after reboot. It didn't save the hardware setting.

The desktop-live isos already have firmware-amd-graphics installed. Maybe the version from backports is needed. I don't have an easy way to test that right now. And debian wiki suggests that this is not the problem. (4.19 is new enough)

Missing from beomint:
https://paste.debian.net/1192135/

Missing from debbowulf:
https://paste.debian.net/1192136/

The one that looked obvious was debian-mate-default-settings, which is missing in the system that does save its settings. So I tried removing that from my mate. The result was no panel and no righ-click menu on the desktop.

I also tried installing and running dconf-editor to see if there was something about audio settings in it. There is. Toggling "use default mixer device" did not fix it.

I don't know what to try next. Do you know where mate saves its settings?

uh-oh...

I haven't been paying close attention to this thread.

live-config messes with anacron via the live-config script, /lib/live/config/1110-anacron which uses dpkg-divert to disable anacron. This is useful in a live-CD where everything is read-only.
This only activates when you boot into a live system, not an installed system.

Refractainstaller copies the RUNNING live system to hard drive.

You just uncovered a 10-year-old bug that I didn't know about. The installer needs to undo this during the installation.

I suspect that the right way to undo it is to use dpkg-divert.

# ls -l /usr/sbin/anacron*
lrwxrwxrwx 1 root root     9 Apr  2  2018 /usr/sbin/anacron -> /bin/true
-rwxr-xr-x 1 root root 38928 Feb  6 14:18 /usr/sbin/anacron.orig.anacron

# dpkg-divert --remove /usr/sbin/anacron
dpkg-divert: warning: please specify --no-rename explicitly, the default will change to --rename in 1.20.x
Removing 'diversion of /usr/sbin/anacron to /usr/sbin/anacron.orig.anacron by live-config'

And then to verify that it really did what it was supposed to do (but did not):

# ls -l /usr/sbin/anacron*
lrwxrwxrwx 1 root root     9 Apr  2  2018 /usr/sbin/anacron -> /bin/true
-rwxr-xr-x 1 root root 38928 Feb  6 14:18 /usr/sbin/anacron.orig.anacron

Yet dpkg-divert thinks it did the right thing:

# dpkg-divert --remove /usr/sbin/anacron
dpkg-divert: warning: please specify --no-rename explicitly, the default will change to --rename in 1.20.x
No diversion 'any diversion of /usr/sbin/anacron', none removed.

Computer, do as I say!

# mv /usr/sbin/anacron.orig.anacron /usr/sbin/anacron
# ls -l /usr/sbin/anacron*
-rwxr-xr-x 1 root root 38928 Feb  6 14:18 /usr/sbin/anacron
root@nomad:/home/phred# 

I'll reboot and see what happens.

Update: I see this in syslog for the first time after reboot. I think it's working now.

Apr  1 15:44:42 localhost anacron[1828]: Will run job `cron.daily' in 5 min.
Apr  1 15:44:42 localhost anacron[1828]: Will run job `cron.weekly' in 10 min.
Apr  1 15:44:42 localhost anacron[1828]: Will run job `cron.monthly' in 15 min.

If you have one of each of these systems, you could compare package lists. Or make two package lists and paste them somewhere like paste.debian.net and I will compare them.

dpkg -l | awk '/^ii/ { print $2 " " $3 }' > package_list

Here's the final version (I think). Tested in xfce (thunar) and mate (caja), both in chimaera. I'll rebuild the package later today.

.rubberband,
.view .rubberband,
view rubberband,
rubberband {
	background-color: alpha (@theme_selected_bg_color, 0.35);
	border-color: @theme_selected_bg_color;
	border-style: solid;
	border-width: 1px;
	border-radius: 2px;
}

On my T420 running chimaera, I get the same message about iwl-debug-yoyo.bin, but there's no boot delay. I don't have any error messages about regulatory.db. That loads ok.

I don't know where to go with this. Just thought I'd add some data points.