adw
February 19, 2025, 12:46pm
13
@marmarek : Should we have some kind of CI-like infrastructure that automatically tests all advertised download methods (e.g., at least the main ISO and torrent links on the downloads page) to ensure they exist and work (e.g., try downloading the first few KB), and only announce the release after this automated system gives us a green light?
barto:
As usual, updating a R4.2.x install, fully, does not result in the system being labelled as R4.2.4 (am writing “as usual”, because it happened with 4.2.3 too, see here )
From the thread you linked:
That would make sense if the one-week delay were intentional, but it seems more like an unintended bug (i.e., someone just forgot to push it to stable). I don’t recall ever hearing that there’s supposed to be a standard, intentional one-week delay, and I’m not aware of it being mentioned in any our written policies (such as the release schedule documentation). Granted, this does seem to be happening pretty often lately, which does suggest it may be intentional.
But if it is intentional, then… …
Forgot this one push, done.
So, it looks like it’s not intended or expected behavior and therefore shouldn’t be in the announcement.
Opened an issue for this:
opened 12:45PM - 19 Feb 25 UTC
P: major
C: tests
### The problem you're addressing (if any)
We often forget to update the versio… n number when issuing a new release:
- https://forum.qubes-os.org/t/fixed-r4-2-3-fully-updated-system-still-shows-4-2-2/29024/
- https://forum.qubes-os.org/t/qubes-users-qubes-os-4-2-4-has-been-released/32354/2
- https://forum.qubes-os.org/t/qubes-os-4-2-4-has-been-released/32355/12
It happens often enough that some users have started to assume it's intentional and blame the announcements instead for not mentioning. I [asked](https://forum.qubes-os.org/t/fixed-r4-2-3-fully-updated-system-still-shows-4-2-2/29024/7) whether it's intentional, and @marmarek [indicated](https://forum.qubes-os.org/t/fixed-r4-2-3-fully-updated-system-still-shows-4-2-2/29024/8) that it was just due to a forgotten push.
### The solution you'd like
Since it's easy to forget to update this, we should add some kind of automated test or CI check to catch this before issuing each new release.
Perhaps even just checking to make sure that the version number has changed at all (or isn't equal to the old one) might be enough of a test.
### The value to a user and who that user might be
Users who do in-place upgrades will be able to verify that they're on the correct release.
### Completion criteria checklist
_No response_
4 Likes