SRU Justification:
Impact: In order to lower maintenance burden we want to use the same code base of ubuntu-fan throughout all releases. Most of the code already is the same. The remaining delta could be split into two classes: documentation (for which there is a separate bug report and changes to build in and/or DEP8 testing (which this report is about).
Fixes:
[ Xenial SRU only]
* debian/control: fix use of obsolete > operator
This only changes a package dependency. Not affecting use.
* fanatic: fix local testing and improve docker test
This fixes up the built-in test for docker and improves it
(check for container start failure, potential clean up before
start). Executed as part of DEP8 testing.
* fanatic: Support nc -N option
Strictly only needed in Artful but since the code was written
to check for the new option and only use it if present, there
is no visible change in previous releases. Covered by DEP8, too.
* fanatic: Prevent error message without docker installed
Without docker installed there is a message to stderr when
disabling fan. This broke the LXD dep8 test. So again, verified
by DEP8 as the LXD part gets added.
* fanatic: Supply LXD storage if needed
At some point the lxd-client started to require providing --storage
when creating containers. The LXD dep8 test would break then. Fix
checks for option being available and conditionally provides it.
Again covered by DEP8 testing.
* dep8: Add LXD auto-test
Adds DEP8 testing for LXD (currently Xenial only tests docker).
[ Xenial and Zesty SRU ]
* DEP8: lxd: Suppress stderr on profile list(*)
Strictly only needed in Xenial but we want to make the code the
same. In Xenial doing a lxc profile list is/was printing a hint
about init to stderr even when lxd init was already done. This
broke the LXD DEP8 test in Xenial. The fix is to ignore stderr
for that call which again is verified by DEP8 passing.
Test case:
As stated for the individual fixes all the changes will be verified by the DEP8 tests passing. The only exception is the packaging fix and that has no impact on usage at all.
Regression Risk:
Since all is covered by DEP8, and only modifies tests, the overall risk of regression for normal operation should be very low.
It's a bit of a special case this bug (and because of that it's also a bit confusing), but I'd still like to see some additional things added to the bug description: is there a way how this can be verified as 'working'? Would be nice if there was a test case introduced, someway that one could decide if the upload indeed has working changes for this task (there would be probably more testing for xenial). I'd also like to see a regression potential analysis of those changes, since for someone from the outside it's hard to guess looking at the diff which parts belong to this bug and how they can affect other parts potentially.