00:39 <isd> Reminder: office hours in 20 min.

02:32 <abliss> ok, i uploaded my pty hack-in-progress to https://github.com/abliss/upty . I'll be shocked if you are able to use it for anything yet, but at least I have a place to push my checkpoints now.

10:34 <CcxWrk> abliss: ewww! Good job though! :D

20:54 <isd> The way humans keep track of time is utterly insane.

20:55 <isd> (context: working on calendar app)

21:29 <abliss> maybe this is the best moment in history to try switching to decimal time again.

21:30 <abliss> schools are closed, tens of millions out of work, fewer scheduling appointments than ever

21:31 <isd> Killing time zones would be sufficient. I'd be okay with everyone being on UTC and folks who are currently in US/Eastern eating lunch around 16:00.

21:32 <isd> That alone would be a massive simplfiication

21:36 <JacobWeisz[m]> Yeah, time zones seem pointless to me.

21:37 <isd> Getting rid of DST would also be a massive simplification

21:38 <isd> Like, DST makes it so that it's actually not even possible to convert from one time zone to another unambiguously.

21:40 <isd> So, I'm very glad the Haskell packages for dealing with this stuff force me to think about what I really want here. I just wish the domain they're modeling wasn't so ridiculous

21:40 <abliss> I used to be anti-timezone, but i've softened over the years. But I still think we'd be much better off without weeks, months, years, hours, minutes, seconds, and of course DST.

21:41 <abliss> For my money wey should call jan 1 2000 the new zero point, and track years+days+decimal days from then (in local time). we can work 7 out of 10 days (say 5 on, 2 off, then 2 on and 1 off)

21:43 <isd> I would be happy if I didn't have to deal with the fact that a date & time stamp in local time isn't actually isomorphic to a specific point in time.

21:44 <isd> Everything else is vaguely annoying conversion math, but that's actually a deeply absurd thing to have to model

21:57 <abliss> here's one of the best defenses of timezones i've seen for nerds who want to redesign timekeeping: https://qntm.org/abolish

22:08 <isd> As someone who's parents go to bed are a 15 minute drive away yet go to bed 7 or 8 hours before I do, I'm not totally convinced.

22:08 <isd> ...but DST is the big problem.

22:10 <isd> I think there's a strong case to be made that most things time zones are good for are themselves dumb societal conventions.

22:11 <isd> I'm not sure I buy it, but I think you can make the case.

22:52 <kentonv> I once wrote some forum software in which I wrote my own handling of American DST rules for rendering timestamps of posts, and then Congress promptly changed the rules the following year (after being stable for decades).

23:08 <abliss> if you were only handling North American DST rules, you had it easy. there are some harrowing stories in the tz database about morocco, turkey, kazakhstan etc

23:11 <abliss> e.g. https://mm.icann.org/pipermail/tz/2018-November/027146.html

23:54 <abliss> golang seems so nice and simple when you're just compiling files in one package, but every time i have to interact with the import/package/module/goroot/gopath mess, i end up crying into fistfuls of my own hair

23:55 <isd> go modules bring much needed sanity to the package management story.

23:55 <abliss> so far all i'm seeing is increased insanity caused by the variety of different and conflicting docs

23:56 <isd> I've basically learned to do go mod init <pkgpath> as soon as I need to import something outside the stdlib. From there things are pretty good

23:57 <abliss> is there a separate magic incantation to make protobuf imports work?

23:57 <isd> Where does pb come into this?

23:59 <abliss> i honestyl don't know where it comes in, but every time i try to make use of any gvisor code, my `go build` commands all fail on https://github.com/google/gvisor/blob/master/pkg/eventchannel/event.go#L29