Comment 5 for bug 1555700

Discussion from #ubuntu-release

<infinity> rharper: As for the AIO thing, as you understand the code, what are the odds this could have a negative impact on users of other backends if the AIO stuff is broken?
<infinity> rharper: If you're 99% confident that the change is meaningless for people who don't use the feature, then I'm fine with adding it.
<rharper> infinity: ok; re aio; let me look a bit to be certain, libaio is standard stuff; the question is has tgt had aio for quite some time and just not enabled it (and why); debian recently added it
<infinity> rharper: Right, it's obviously been there longer upstream than in Debian, the question is how long and, more importantly, is it isolated enough to not blow up the world for non-AIO users if it sucks.
<infinity> rharper: The latter question being the interesting one. I don't care if the feature it experimental and added last week if it has zero impact on people who don't use it.
<infinity> s/feature it/feature is/
<rharper> understood
<infinity> rharper: Feel free to copy and paste any of the above to the bug, FWIW, and if your conclusion is "yeah, it looks safe enough", no need to round-trip to me again, FFe approved if your best judgement says it's safe.
<rharper> infinity: will do; I need to confirm aio is only enabled via user choice (ie not default); if so then I'd say it's out of the way for anyone except those whom request it;
<infinity> rharper: And if you're less sure, a commitment of "we'll actively watch for explosions and either fix the issues or revert the aio addition", then also approved. :P
<infinity> rharper: With a caveat that "revert the aio addition" can only be a solution pre-release. Once xenial is out the door, you're stuck supporting whatever features we released with.
<infinity> rharper: If you're okay with the above, go nuts.