<abathur>
LnL: is there anything else I need to button up on the PR?
<LnL>
gchristensen: ^ did you get any new information?
<gchristensen>
I heard fairly consistent feedback that big companies had rules against datasets without filevault
<gchristensen>
I think it is okay to make this informational and do it anyway
<gchristensen>
if someone needs it that bad, they can send a PR
<LnL>
yeah, the warning is there now
<gchristensen>
cool
<gchristensen>
I think it isn't wonderful, and maybe we'll do something different next release
<gchristensen>
but let's do this for now
<LnL>
only question for me is if we should keep unencrypted in the flag, it's a bit overly pessimistic but does set better expectations for the company case
<gchristensen>
maybe something like without-filevault ?
<gchristensen>
more specific, and it still does not install if there is no t2, right?
<LnL>
yeah the conditons are good, just the name I'm not sure about
<gchristensen>
yeah
<gchristensen>
yeah so it isn't that it is completely unencrypted, it just doesn't have filevault
<gchristensen>
(weakly held opinion)
<LnL>
abathur: you got any thoughts on this?
<gchristensen>
my strongly held opinion is we shouldn't scare users unnecessarily, and we shouldn't pretend like things are totally fine and let users easily poke their eye out. there is a ton of room in there :P
<abathur>
nod
<abathur>
not really sure, it is a bit of a lie, but my ideal case is, you know, that they have to have the implications cross through their brain as they type it out
<gchristensen>
which is the lie?
<abathur>
rather than being able to, justifiably or not, say they didn't know what yes/no filevault means
<abathur>
that it is unencrypted when it may be encrypted-at-rest depending on setup
<gchristensen>
I thought we established in the case where there is no T2 chip, but other encyrpted datasets existed, it would absolutely refuse to install
<abathur>
I guess that's right; I'm also lumping in the old-hardware cases but they won't be using the flag
<gchristensen>
so passing this flag is only saying, if there is a T2 chip, and some datasets are encrypted with filevault, create that dataset without filevault. and it applies in no other case, right?
<abathur>
well, maybe LnL has a thought here; at first I locked the flag so it wouldn't even run pre-catalina
<abathur>
but he did suggest letting pre-catalina run it if they wanted
<gchristensen>
gotta run for a bit
__monty__ has quit [Quit: leaving]
<LnL>
that doesn't change this however
<abathur>
but it might somewhat affect what the flag "means"
<LnL>
more likely to be old hardware, but the conditions stay the same
<abathur>
if filevault is in use, it only creates the volume if a t2 chip is there; if filevault is *not* in use, it will just create the volume
<abathur>
(and it'll end up either unencrypted on old harware, or encrypted-at-rest on new hardware)
<abathur>
so saying without-filevault or non-filevault seems correct
<LnL>
yeah
<abathur>
but I have less of a sense of how clear it is (what fraction of potential Nix installers on a macOS system will type --create-nix-store-volume-without-filevault
<abathur>
and understand how to unpack what that means)
<abathur>
vs the fraction who'll run it without understanding
<abathur>
it might be a non-issue, I won't object if the general perception is that the kind of person interested in Nix will understand
<LnL>
somebody who doesn't know what filevault is probably won't have it enabled
<LnL>
and it doesn't allow the bad case, just doesn't cover more restrictive policies