#kspacademia on 2018-07-05 — irc logs at esper.irclog.whitequark.org

00:17 <egg|zzz|egg> !wpn bofh

00:17 <egg|zzz|egg> !wpn UmbralRaptop

00:23 <bofh> !wpn egg|zzz|egg

00:24 <UmbralRaptop> Did Qboid give bofh a <egg|future|egg>

00:26 <bofh> LOL

02:44 <SilverFox> https://youtu.be/u_xQ_C5RUfo

02:44 <kmath> YouTube - How to make a Klein Bottle from an old pair of jeans - Numberphile

02:44 <SilverFox> yall maths are crazy

08:43 <egg|zzz|egg> TIL https://en.wikipedia.org/wiki/Belarusian_ruble#Third_ruble

08:43 <egg|zzz|egg> each with its own ISO 4217 code

10:51 <UmbralRaptop> <@soundnfury> rqou: reading maths (or "majoring in math", as you call it) is definitely the best thing <@soundnfury> if you can survive a maths degree without going insane, you can excel in any other career <@soundnfury> and if you _do_ go insane, you can become a mathematician! ;)

10:52 <UmbralRaptop> bah, no ☽🦊

11:01 <bofh> I mean, they're not *wrong*

13:17 <egg|work|egg> !meow

13:17 <egg|work|egg> hmmm

13:21 <egg|work|egg> !meow whitequark

13:22 <UmbralRaptop> !meow 55 Cnc e.

14:09 <egg|work|egg> !wpn bofh

14:17 <bofh> !wpn egg|work|egg

14:17 <bofh> a unit*ary* cell, as opposed to a unit cell? interesting.

14:20 <egg|work|egg> bofh: huh, apparently transconjugant exists as a biology term in english, whereas transconjugate doesn't for conjugate transpose (french has transconjuguée)

14:37 <UmbralRaptop> advertised post-doc: 0 < required experience ≤ 4 years

14:37 <UmbralRaptop> 😬

14:58 <egg|work|egg> !wpn UmbralRaptop

15:08 <egg|work|egg> snek

15:38 <egg|work|egg> !u 🛪🛦🛫✈🛨🛩🛧🛬

15:38 <Qboid> U+1F6EA NORTHEAST-POINTING AIRPLANE (🛪)

15:38 <Qboid> U+1F6E6 UP-POINTING MILITARY AIRPLANE (🛦)

15:38 <Qboid> U+1F6EB AIRPLANE DEPARTURE (🛫)

15:38 <Qboid> U+2708 AIRPLANE (✈)

15:38 <Qboid> U+1F6E8 UP-POINTING SMALL AIRPLANE (🛨)

15:38 <Qboid> U+1F6E9 SMALL AIRPLANE (🛩)

15:38 <Qboid> U+1F6E7 UP-POINTING AIRPLANE (🛧)

15:38 <Qboid> U+1F6EC AIRPLANE ARRIVING (🛬)

15:39 <egg|work|egg> !wpn bofh

15:47 <bofh> !wpn egg|work|egg

15:47 <bofh> UmbralRaptop: what the hell, why is the astro job market even worse than normal?

15:48 <UmbralRaptop> ¯\_(ツ)_/¯

15:51 <UmbralRaptop> Imperial University has a rep for being a bad place to work at, right?

15:55 <egg|work|egg> bofh: neat thing in Normandy: https://goo.gl/maps/4oFUnXD7B672

15:55 <bofh> what on *earth* is that rock in the middle of the image just kinda... standing there?

15:56 <egg|work|egg> that's no rock

15:56 <bofh> it looks like a different type of basalt than the cliffs too

15:56 <bofh> oh?

15:56 <egg|work|egg> it's a german bunker from the atlantic wall

15:56 <egg|work|egg> but they didn't build it like that obviously :-p

15:57 <bofh> Okay, true. :P

15:57 <bofh> And that makes sense.

15:57 <egg|work|egg> what happened is that the eroding cliff was reaching where the bunker was built (it was built on the cliff for obvious tactical reasons)

15:57 <egg|work|egg> so the fire dept blew up a bit of cliff, and it fell that way

15:58 <bofh> Oops.

15:58 <egg|work|egg> that was in the 90s I think, the cliff since kept eroding, and now there's a bunker in the middle of nowhere

16:00 <bofh> Does the bunker go into the ground at all, or is it just sitting on the ground effectively?

16:01 <egg|work|egg> it seems to be propped up by some stones

16:02 <egg|work|egg> okay https://qdehais.photoshelter.com/image/I0000QZOu._GbdcU says 1995

16:02 <egg|work|egg> that cliff erodes fast

16:03 <egg|work|egg> bofh: I mean it's a cuboid, so you can probably see how much of it is buried

16:04 <bofh> ahh.

16:04 <bofh> also christ that is some insane erosion rate there.

16:05 <egg|work|egg> bofh: where do you think all the pebbles come from

16:06 <bofh> Well, yes, but that's, like, 10-20m in a 23-year timespan.

16:08 <egg|work|egg> bofh: makes for some vast expanses of complete mess at low tide: https://goo.gl/maps/MvAZK4Rcq232

16:15 <egg|work|egg> bofh: amusingly, the cliff noticeably protrudes out to sea at the level of the bunker if you look on a satellite photo, because http://img.over-blog-kiwi.com/0/53/49/10/201309/ob_6843510d2f20a6c8520831968cf6ccb7_39.jpg

16:17 <bofh> I mean that makes sense given its construction

16:21 <bofh> Huh, that's a *lot* of algal mats at low tide there.

17:33 <egg> meow

17:37 <egg> !wpn whitequark

18:25 <egg> UmbralRaptop: bofh: follow up on the [guinea fowl] vs. [pintade] image search results: try [quail] vs. [caille]

18:55 <bofh> egg: LOL

19:03 <bofh> UmbralRaptop: in its defense, it's not usually this dry

19:03 <bofh> (like seriously I think the Midwest is rainier this year than, say, London. WTF?)

19:04 <UmbralRaptop> Iä!

19:15 <UmbralRaptop> !pet egg

19:16 <UmbralRaptop> Fun things: being sufficiently worn out mentally and physically that you need to hide somewhere and sleep, but you really should socialize more.

19:18 <egg> !pet UmbralRaptop

19:23 <bofh> !pet UmbralRaptop

19:25 <UmbralRaptop> !pet bofh

19:32 <UmbralRaptop> Optimistically: use 3-5 filters to say cloudy/clear, and if clear some vague properties (mass, temperature, metallicity). https://twitter.com/Exo2Conf/status/1014895366314086402

19:32 <kmath> <Exo2Conf> Natasha Batalha: ⏎ 'The Intricacies of Using Color-Color Diagrams to Classify Directly Imaged Planets' ⏎ #Exoplanets2 https://t.co/FLdMF1rMPm

19:32 <UmbralRaptop> Pessamistically, you get murdered by clouds. Just like amateur astronomy!

19:38 <bofh> And now that more and more S/Cs are using Ka-band downlinks, just like deep space communication! :P

19:39 <bofh> (in reality you switch to X-band at that point and store the stuff you need a high-b/w link to downlink for later on the S/C).

19:41 <UmbralRaptop> hah

19:57 <UmbralRaptop> Yay, no leap second. http://hpiers.obspm.fr/eop-pc/index_mobile.php

20:03 <egg> UmbralRaptop: oh nice, thanks https://hpiers.obspm.fr/iers/bul/bulc/bulletinc.56

20:03 <egg> I need to make a pull request then,

20:04 <UmbralRaptop> via, uh, iximeow?

20:18 <bofh> https://quantumfrontiers.com/2018/06/24/whats-the-worst-that-could-happen/

20:28 <iximeow> the leap second thing! woo

20:38 <bofh> no leap second is good leap second.

20:39 <egg> bofh: hey leap seconds are good

20:39 <egg> alternatively, find a way to stabilize the earth's rotation >_>

20:40 <bofh> Alternatively, force the ITU to adopt TAI like a reasonable broadcast institution.

20:41 <bofh> I have no issue with leap seconds if they're being used in astronomy :P

20:42 <egg> bofh: wat, you don't want civil time to shift away from UT1

20:42 <egg> bofh: the stupid thing is making computer systems run on civil time, not having civil time be UTC-based

20:43 <bofh> So I'm complaining about computer system time, not civil time.

20:43 <bofh> Civil time can run on whatever time system it wants

20:43 <SnoopJeDi> the solution is obvious: get rid of the users and all of their messy "needs" will go away

20:43 <egg> sure, and that's what UTC is for

20:44 <egg> bofh: then silly people run their computer system on UTC, and live with the consequences :-p

20:44 <bofh> Yeah, but I'm talking about the ITU, i.e. broadcast TV transmission time synchronization.

20:44 <bofh> People run their computer systems on UTC b/c the ITU uses UTC, that's the chain of events.

20:44 <bofh> And the ITU uses UTC b/c ???

20:44 <egg> it seemed like a good idea at the time :_p

20:44 <bofh> I mean that's the only answer I'd believe, so :P

20:45 <SnoopJeDi> also the explanation for most datetime problems

20:45 <egg> (including python datetime)

20:45 <SnoopJeDi> egg, what do you have against the datetime library?

20:46 <egg> I work with it and my job involves showing times to users in various timezones from various other timezones :-p

20:46 <SnoopJeDi> and...

20:46 <egg> the only good time representation is astropy.time.Time :-p

20:57 <egg> essentially the representation as a calendar date and a time rarely useful except at the point where it is show to someone (at which point you go through ICU and use ICU's calendar types anyway), and tends to induce a lot of confusion when you just want something that represents instants

20:57 <egg> compare astropy.time.Time which just represents an instant hence my comment :-p

20:58 <egg> (or more practically, absl::Time which also just represents an instant but doesn't bother with timescale considerations which are mostly irrelevant outside of astronomy)

21:01 <egg> !wpn bofh

21:04 <bofh> Fair. Meanwhile I'm usually only either interested in relative time, or in precise frequency.

21:05 <bofh> And not even what the values of the relative time is, just that each interval is the same. So, basically what I actually care about is frequency references. Unsurprisingly.

21:12 <egg> bofh: yeah then you want atomic time obviously

21:12 <SnoopJeDi> nothing about time is obvious, ever

21:13 <bofh> !pet the FE-5860A on my desk

21:13 <bofh> egg: :P

21:13 <egg> exactly

21:13 <egg> s/ex/eggs/

21:13 <Qboid> egg meant to say: eggsactly

21:14 <egg> bofh: but that only works if you want a local frequency standard, otherwise you should run on actual TAI,

21:14 <egg> bofh: or if a post hoc frequency standard can work then TT(BIPM)

21:15 <bofh> True, but you *can* entrain an atomic reference to TAI in the same way that GPSDOs entrain an OXCO to TAI.

21:16 <bofh> Which has the benefit that it is actual TAI, but also if you lose internet and ICE drives by with a GPS jammer, you still have a low-drift reference that works :P

21:16 <SnoopJeDi> ooph

21:16 <egg> bofh: of course

21:17 <bofh> (I wish I were joking about the ICE thing, https://www.kg4cyx.net/department-of-asshole-blockades-dab-the-random-border-patrol-check/ )

21:17 <egg> bofh: and then you can still convert everything back to UTC and derived timezones for display, everyone is happy :-p

21:17 <bofh> Exactly! :P

21:17 <egg> yeah saw that

21:18 <SnoopJeDi> wtf am I reading

21:19 <egg> bofh: I mean using UTC as a representation has basically the same kind of issue as using a calendar as a representation, it confuses representation used to display an instant with a simple stable instant representation with good properties

21:19 <egg> it just has those issues less often since at least you have avoided the timezones

21:20 <bofh> SnoopJeDi: friend bought a surplus TV news studio camera, was driving back home, went to go pump gas, got stopped by an ICE goon who noticed the camera, proceeded to hide asking if it was currently on studio uplink, upon hearing "yes" he decided to GTFO.

21:21 <bofh> which, like, inspires a lot of confidence, between the cellphone jammer and the hiding-identity-and-leaving upon realizing that they're being recorded by a non-cell-uplink camera.

21:21 <bofh> egg: true

21:22 <SnoopJeDi> oh, you know this person?

21:23 <bofh> SnoopJeDi: yeah

21:23 <SnoopJeDi> okay that puts a touch more credence to this then

21:23 <egg> bofh: an amusing thing is that for many applications, if you use any smooth timescale that doesn't diverge from UTC/UT1, you can largely ignore the leap seconds even when converting back to civil time for rendering, since you rarely render seconds near the leap anyway

21:24 <bofh> egg: yeah, like that's another reason why computer time having leapseconds bothers me. they're rarely *relevant* in the case you do want UTC/UT1, namely display.

21:24 <egg> well, you don't want the drift though

21:24 <egg> because the leap seconds do add up

21:25 <egg> but basically a computer system that somehow uses UT1 for timestamps can naively render assuming it uses UTC for most purposes (not all), and has all the good monotonicity/subtractibility properties

21:25 <bofh> More or less.

21:26 <egg> (obviously you don't want to actually use UT1 because then your clocks are replaced by observatories, but e.g. https://developers.google.com/time/smear)

21:26 <bofh> rofl

21:26 <SnoopJeDi> o.O

21:27 <egg> the cleaner alternative would be to just store TAI and have the UTC leaps in the rendering layer, but that's somewhat impractical compatibility-wise, here most of the time "I can't believe it's not UTC" works

21:31 <bofh> egg: I mean that's exactly what *should* be the case and what I wish were the case.

21:31 <egg> yes, but legacy :-p

21:33 <egg> bofh: if you develop your own somewhat-self-contained system, by all means use TT or TAI as the time format, and convert for interfacing/display

21:33 <egg> bofh: and, well, that's what happens if the system is designed by astronomers *points at astropy* :-p

21:34 <egg> bofh: similarly principia uses TT everywhere :D

21:34 <egg> (the internal format is seconds of TT from J2000, either binary64 or two nonoverlapping binary64s)

21:34 <egg> !wpn Deille

21:35 <egg> bofh: https://logs.tmsp.io/kspacademia/ is back up!!!

21:35 <bofh> wait, as in double-precision float?

21:35 <egg> and Thomas doesn't really know how :-p

21:35 <egg> bofh: yeah we have DoublePrecision<Instant>

21:36 <bofh> I suppose the precision of doubles is enough that it doesn't matter to use that over 2xuint64_t's or whatnot.

21:39 <egg> bofh: yeah, that's what astropy (or rather SOFA/ERFA beneath) does too

21:39 <egg> it's just so much more convenient to do arithmetic on durations afterwards

21:39 <egg> absl::Time uses integers otoh, because there you don't expect to do much quantity arithmetic

21:41 <bofh> True, if you want to do arithmetic on J2000 timestamps stored as 2xuint64_t's you basically want to overload some operators or it rapidly gets very horrible.

21:43 <egg> bofh: well, whatever you do you want overloaded operators (absl::Time has differences), it's just that if your durations are fixed point and then you do physics with them it gets messy

21:43 <egg> bofh: also a nice bonus of floating-point time is that you get the NaNs and infinities for free, absl::Time has to reinvent infinities

21:44 <egg> (turns out infinite timestamps are *really useful* :-p)

21:44 <bofh> Do you *want* NaNs in your time representation tho? Infinities are useful, but NaNs?

21:44 <egg> well, I don't *want* NaNs, but if I subtract infinities I might as well get that rather than an infinity like in absl :-p

21:45 <egg> (and you just get the whole arithmetic model for free, infinite durations too etc.)

21:45 <bofh> (point, that much is very nice).

21:46 <egg> (absl also has infinite durations obviously)

21:47 <egg> and absl doesn't care about having a concept of frequency, so you don't need to worry about the multiplicative parts of IEEE 754 infinite arithmetic

21:48 <egg> but yeah, fixed-point is always surprisingly hard

21:52 <bofh> Yep.

21:52 <bofh> It's why I absolutely hate writing fixed-point audio codecs, it turns into an unholy nightmare.

21:53 <egg> bofh: Ada has fixed-point types natively!

21:53 <bofh> https://twitter.com/BizarreVictoria/status/1014834700102270976 rofl

21:53 <kmath> <BizarreVictoria> Guys, I just discovered a 13th-century translator named Hermannus Alemannus, or, literally, 'Herman the German'. ⏎ A… https://t.co/Ivg3okSAhz

21:54 <egg> somewhat underspecified though

21:54 <egg> http://www.adaic.org/resources/add_content/standards/05aarm/html/AA-3-5-9.html

21:55 <egg> I like the "To be honest" sections of the AARM :-p

21:56 <bofh> it's always amusing to read those compared to the language of the C++ standard.

21:57 <egg> ah actually "the small may be specified by an attribute_definition_clause (see 13.3);" so you do have fairly fine control if the compiler wills it

21:58 <egg> bofh: there's also a non-annotated version without those random comments, but even then I find the style more pleasant

21:58 <bofh> Yeah, the annotation is a feature, not a bug.

21:58 <egg> bofh: unannotated version: http://www.adaic.org/resources/add_content/standards/05rm/html/RM-3-5-9.html

21:59 <egg> bofh: no I mean even the unnanotated prose of WG9 is more readable than that of WG21 imo

21:59 <bofh> wait, relly?

21:59 <bofh> really*

21:59 <bofh> that's, uh, impressive.

22:00 <egg> caveat: I am slightly biased :-p

22:01 <egg> bofh: admittedly there are weirdnesses (although since Ada 95 the "no other effect" being a defined term has disappeared :-p)

22:08 <bofh> I assume that's synonymous with "no side-effects"?

22:13 <egg> bofh: http://www.adaic.org/resources/add_content/standards/05aarm/html/AA-3-1.html (12.o)

22:16 <egg> bofh: example of the weirdness in RM83: http://archive.adaic.com/standards/83lrm/html/lrm-06-01.html#6.1 "The elaboration of a formal part has no other effect." definition in http://archive.adaic.com/standards/83lrm/html/lrm-03-01.html#3.1

22:54 <egg> cat. https://twitter.com/whitequark/status/695209341205487616

22:54 <kmath> <whitequark> #wikifinds (seriously) https://t.co/KgubEbUxqy

23:18 <hillexed> hey everybody did you know that if you have a list n items long the number of possible permutations you can make out of it by only pushing and popping to and from a stack is equal to the nth catalan number

23:19 <UmbralRaptop> Summoning egg?

23:19 <SnoopJeDi> A brilliant entrance

23:20 <hillexed> summoning eggs sound like a !wpn

23:20 <SnoopJeDi> !wpn -add:adj summoning

23:20 <Qboid> SnoopJeDi: Adjective added!

23:20 <SnoopJeDi> right you are

23:21 <hillexed> !wpn SnoopJeDi

23:21 <SnoopJeDi> !wpn hillexed

23:21 <SnoopJeDi> that's...apropos.

23:21 <UmbralRaptop> !wpn hillexed && SnoopJeDi

23:21 <SnoopJeDi> !wpn UmbralRaptop

23:21 <SnoopJeDi> :o

23:21 <UmbralRaptop> hm

23:44 <egg> hillexed: the number of possible permutations you can make out of it by only pushing and popping to and from a stack << I'm not sure I understand

23:46 <hillexed> It's the number of total arrays you can make if you start with input = [1,2,3...n] and your only options are stack.push(input.pop()) or output.push(stack.pop())

23:56 <egg|zzz|egg> ah yes it bijects to well-formed bracket expressions