Ulrich Müller

> I fiddled around with it a little bit by sourcing `phase-function.sh` (where `pipestatus` is currently defined). This results in a lot of noise because `phase-functions.sh` really expect to be...

Please use "HPND" which is close enough. We know that there are dozens of variants around, where only one or two words differ (but without any difference in meaning). Adding...

Thinking about it, we could of course add a line `HPND-sell-variant = HPND` to metadata/license-mapping.conf. However, that shouldn't be a blocker for this PR.

The spec doesn't say what should happen when a directory belongs to two packages, and these packages specify different owners and modes, and I suppose there is no good way...

Is there an actual filesystem check implementation which does that? Given that 0xfff etc. are legal values (and are the default in both Windows and Linux) I find this somewhat...

So what you're saying is that if the filesystem is filled up with one-cluster files (and the media type in the first reserved cluster is also 0xff), then the FAT...

Some checks fail, but looks like this is a problem with the bash version of the runner (ancient Ubuntu) being outdated, not a problem with the code.

> There's an irony that macos passed this considering their age. Your change is correct but needs to be held off on merge until the test runners can support it.....