make debian broken on master and dunfell

[humanoid2050]

The build scripts do not actually copy the updated image to the toFlash directory.

[htot]

True. Because I didn't want to pollute the toFlash directory while generating the btrfs subvolume. But probably as long as we support also ext4 it makes sense to behave as the original.

BTW: Have you tried the btrfs image? It allows you to dual boot Yocto/Debian.

[htot]

https://github.com/htot/meta-intel-edison

[humanoid2050]

I see that you enabled the cp in your branch. I did the same in mine.

I did try the brtfs image.I didn't know that it was easy to dualboot yocto and debian though. That's a neat capability.

Even though I like some of the btrfs capabilities, they don't really buy me much. I prefer having a bit more control over where my data physically ends up. I also know ext4 and it's comfortable. I also don't really need OTA updates for my use case.

If anything, I'd actually like to set up my Edison to be completely based on the SD card, and just flash images to a card the way I do with the other 15 or so embedded computers in my collection.

Has any thought been given to being able to support btrfs/ext4, emmc/uSD, separate boot/home, yocto/debian and various combinations of those options in a robust way? I know it represents a lot of work, but I think it might be something that could be implemented once and then used for a long time if done right. I wonder if cmake could be leveraged (with its build options and custom commands) to generate the make file dynamically, then generate an appropriately configured set of images and flash scripts.

[htot]

If anything, I'd actually like to set up my Edison to be completely based on the SD card, and just flash images to a card the way I do with the other 15 or so embedded computers in my collection.

You can already do that now, as U-Boot can read the SD card to load the kernel.

Has any thought been given to being able to support btrfs/ext4, emmc/uSD, separate boot/home, yocto/debian and various combinations of those options in a robust way?

Yes, this is already the case. But maybe not all combinations have been tested rigidly.