The issue is that stable has pretty much become synonymous with “previous testing” in Manjaro ARM.
How things should work is that:
- things enter
testing, - regressions are found there by testers (which is already the first issue we are having, e.g., I have not seen a single report of this regression in the previous stable update in
unstableortesting, it was first found instableby me, and had to be fixed in all 3 branches after my report), - regressions are quickly fixed in
testing(which is another issue, e.g., for the current Chromium breakage, the issue was just punted to ALARM), and - once the known regressions are fixed, the update moves to
stable.
How things actually work is that:
- things enter
testing, - several issues stay unnoticed due to lack of testers (see above),
- those that are noticed are sometimes not fixed for weeks due to lack of packagers (e.g. the current one),
- then at some point a new
testingupdate is ready, so at that point, in order to make room for it intesting, the currenttestingis flushed tostable, whether it actually works or not.
This kinda defeats the point of stable.