ec changed the topic of #elliottcable to: a π•―π–Šπ–“ 𝖔𝖋 π•―π–Šπ–™π–Šπ–—π–’π–Žπ–“π–Šπ–‰ π•―π–†π–’π–˜π–Šπ–‘π–˜ slash sΝ”ΜžuΝ•Ν™pΝ™Ν“e̜̺rΜΌΜ¦i̼̜oΜ–Μ¬rΜ™Μ™ c̝͉α»₯Μ§Ν˜αΈ·Μ‘Ν™Ε£Ν“Μ€ || #ELLIOTTCABLE is not about ELLIOTTCABLE
kier has joined #elliottcable
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 3 new commits to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/compare/a8d6506523be...7a4888466b89
<purr\Paws> giraphe/hello-travis 7a48884 ELLIOTTCABLE: + supporting old npms that don't flatten
<purr\Paws> giraphe/hello-travis 70f861d ELLIOTTCABLE: + abandoning non-stable versions
<purr\Paws> giraphe/hello-travis 81cc4b4 ELLIOTTCABLE: + IRC notifications, someday
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/commit/45603e16adae49b5715d9d14bcedbb061cd991ef
<purr\Paws> giraphe/hello-travis 45603e1 ELLIOTTCABLE: + nope that didn't work
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 2 new commits to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/compare/45603e16adae...eb7a3c9fe8fb
<purr\Paws> giraphe/hello-travis eb7a3c9 ELLIOTTCABLE: + install codecov only on CI
<purr\Paws> giraphe/hello-travis 351d2f0 ELLIOTTCABLE: Revert "+ abandoning non-stable versions"...
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/commit/025cf192be4222f8d0cd7f09ee752a0b0cc1f5bf
<purr\Paws> giraphe/hello-travis 025cf19 ELLIOTTCABLE: + switching to new versions of node to transpile
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/commit/c7694618dc59854ea6ade336c6712102c4e1b737
<purr\Paws> giraphe/hello-travis c769461 ELLIOTTCABLE: + oops
<Pyrrh> purr stahp
<Pyrrh> ec what are you doing to your poor bot
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to hello-travis: https://github.com/ELLIOTTCABLE/giraphe/commit/e2aa1bc1e367a2bad7265d839a2754a7e47290ad
<purr\Paws> giraphe/hello-travis e2aa1bc ELLIOTTCABLE: + abandoning old versions again, fuck it
<pikajude> activity? in *my* #elliottcable
_whitelogger has joined #elliottcable
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to Current: https://github.com/ELLIOTTCABLE/giraphe/commit/6a771c5f93c0b6ae0ea16f73d03bd5d7179601d7
<purr\Paws> giraphe/Current 6a771c5 ELLIOTTCABLE: (meta fix cov) Fix Travis and Codecov CI...
<elliottcable> pikajude: shush
<elliottcable> oh wow bot
<elliottcable> jesus christ
<elliottcable> was trying to make travis-ci work
<elliottcable> I usually have such trash-branches named β€œblah+”, and then don't notify IRC on trash-branches
<elliottcable> mis-named that branch apparently, oops
<elliottcable> ljharb: ur so helpful omg
<elliottcable> Pyrrh: hello
<pikajude> elliottcable:
<elliottcable> gonna be some more noise sry
<elliottcable> fuck this conference has been a clusterfuck
<elliottcable> my sleep-schedule is so far of
<elliottcable> off
<elliottcable> out of, uh, sun/mon/tue/wed, four days of conference, I've actually managed to *attend* 1.5 days, mon and half of tue
<elliottcable> my flights were all delayed massively, resulting in me actually arriving in Barcelona at like 7am on Sunday, running on no sleep β€” so I obviously missed everything that day ...
<elliottcable> then jet lag and insomnia have basically had a beautiful, perfect, Satan-baby, leaving me awake all night every night and sleep-deprived as fuck every day
<elliottcable> the material at this conference is *way* over my head, like massively, and would be difficult enough if I weren't trying to keep up on a goddamn cocktail of stimulants
<elliottcable> and none of that is to mention the 35Β° / 80% humidity fucking weather
<purr\Paws> [giraphe] ELLIOTTCABLE force-pushed Current+ from 63fa1c6 to 69cf238: https://github.com/ELLIOTTCABLE/giraphe/commits/Current+
<purr\Paws> giraphe/Current+ cc67060 ELLIOTTCABLE: + fixing a no-op test
<purr\Paws> giraphe/Current+ 77f9e2d ELLIOTTCABLE: + improving test-permutation function, and adding edge-option permutations
<purr\Paws> giraphe/Current+ ca0835c ELLIOTTCABLE: + documenting forthcoming func
<joepie91> either my dep tree deduplication code is broken
<joepie91> or the code that's supposed to verify it, is
<ljharb> elliottcable: lol how so
<ljharb> oh you mean my tmbg reference
<elliottcable> ljharb: :P
<joepie91> excellent, it's a bug in my validation code
<joepie91> not in the implementation
<joepie91> \o/
_whitelogger has joined #elliottcable
<elliottcable> god, Paws is dumb
<elliottcable> somebody help me
<elliottcable> make me stop
<elliottcable> make me stop caring
<elliottcable> make me stop working on it
mylesborins has quit [Quit: farewell for now]
mylesborins has joined #elliottcable
<Cheery> elliottcable: why do you care about it if you think it's .. um..
<Cheery> nevermind. carry on.
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 1 new commit to Current: https://github.com/ELLIOTTCABLE/giraphe/commit/c0c7307288456688b1d8e2db1f3d095b25ade1c5
<purr\Paws> giraphe/Current c0c7307 ELLIOTTCABLE: (- meta pkg) Support npm@5...
<elliottcable> Cheery: hahaha exactly ;_;
<elliottcable> Cheery: idk I'm obsessed or mentally ill I think who knows probably something
<elliottcable> Cheery: you're designing a language too, yah?
<Cheery> yup
<elliottcable> I'm at PLDI right now!
<elliottcable> it's so far beyond me
<elliottcable> here's hoping you have more idea what you're doing than I do ;_;
<Cheery> the two last days I've been diverging to writing scripts that fetch new reddit posts from my favorite subreddits and does naive bayes on their title.
<Cheery> though I'm doing that in python
<Cheery> results were encouraging so I'm studying some more document classification stuff.
<Cheery> I hope to produce a "secretary" to myself that filters out all the utter crap from the mess that's internet.
<elliottcable> lolololol
<elliottcable> reinforcing your bubble programmatically sounds like a huge waste of effort :P
<Cheery> I'm not thinking about a bubble, I'm more like thinking about noise vs. I've-seen-that-before vs. signal.
<Cheery> also I can re-introduce some crap into it with random generator.
<elliottcable> idk, for me, Reddit is 97% noise
<elliottcable> I stay mostly th'fuck away
<Cheery> yup. But there are the gems.
<elliottcable> I get a lot more value from Twitter, due to a carefully-curated following
<elliottcable> … of 2,000+ people, but still 🀣
<Cheery> I should probably prune my twitter though.. :)
<elliottcable> Cheery: come write Pratchett w/ me!
<elliottcable> ;)
<elliottcable> -didja
<elliottcable> -didja didja
<elliottcable> I miss purr
<Cheery> ? wut
<elliottcable> hahaha nothing'
<elliottcable> before your time
<elliottcable> I used to spend a weirdly inordinate amount of my time trying to convince people to write implementations of Paws for m
<Cheery> I've thought about writing interactive fiction sample program in Lever.
<elliottcable> it's supposed to be a small language at its core, if you ignore a few of the concurrency constructs kinda/sorta β€” think Scheme
<Cheery> to illustrate the parser.
<elliottcable> so theoretically people were supposed to be able to knock out a simple 90%-implementation of Paws in, like, a weekend, in their favourite language
<elliottcable> problem is I'm really bad at explaining my work v.v
<Cheery> I'm thinking about a dracula castle -story, but I keep hitting storylines where it diverges too much.
<elliottcable> lmao
<elliottcable> wait what's that got to do with a parser
<Cheery> "walk to garden shed" "kick the coffin"
<Cheery> not necessarily in that older.
<Cheery> order*
<Cheery> I've also thought about writing a dartmouth basic with the same thing..
<Cheery> so I suppose if Paws is a weekend -thing, you could probably get me to write an implementation of it in Lever.
<Cheery> stuff is accumulating though. :)
<Cheery> 15:22 <+elliottcable> reinforcing your bubble programmatically sounds like a huge waste of effort :P
<Cheery> this is actually quite serious concern.
<Cheery> I think what I want to do is kick off the crap that just wastes my time.. The kind of gambling elements that catch your interest but often turn out to be empty.
<Cheery> except the ones that resemble those that paid off in the past.
<Cheery> Then I also want to separate out the kind of overflowing self-promotion, and of course all the guys who yell at the wolf to gain attention.
<Cheery> it may be that's 10% of stuff in the end.
<Cheery> but it's easily the sort of stuff that distracts and consumes time from all the good stuff.
<Cheery> Then there's the opposite.. I want it to highlight the kind of stuff with very high information value.
<Cheery> and stuff that's interesting to me.
<Cheery> and for the rest I sort of want it to just pass it plain.
<Cheery> maybe some sort of sorting by signal/noise, independent of what's the signal.
<Cheery> But for it to do that, it needs a good classification model.
<Cheery> I think titles won't do it all the way through.
<Cheery> anyway.. you get the idea.
<elliottcable> so what's Lever
<elliottcable> bear in mind I have low band-width
<elliottcable> is it typed, strict, static, by any stretch of the imagination? if so, I have no bandwidth for it
<elliottcable> (not in a critical way just a β€œthings I care about” way!)
<elliottcable> if not ... gimme the, uh, four-ish-line elevator-pitch, b/cuz I don't remember whatever you've told me in the past 🀣
<Cheery> it's dynamically typed.
<Cheery> ok. I'll tryr.
<elliottcable> :3
<elliottcable> and forgive lack of an immediate reply, too β€” I'm sitting here trying to digest even 1/10th of a talk about tree-flattening-operation-compilation in a polyhedral compiler
<elliottcable> i.e. Shit I Don't Grasp At All
<Cheery> where?
<Cheery> ohwait.. there.
<elliottcable> Barcelona
<Cheery> yeah.
<elliottcable> pldi17.sigplan.org
<elliottcable> I can't load that page but presumably it says something useful ;_;
<Cheery> shows out funny looking buildings, spotted out few big brands that I suppose are sponsors, bunch of faces of guys I don't know..
<Cheery> oh there's the program of the day.
<elliottcable> here's the entire programme
<Cheery> it's multiple-day event
<elliottcable> but it's, god, ... 1, 2, 3, 4, 5m 6m 7m 8m 9m 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23?
<Cheery> and this doesn't show the current time. But I see the polyhedral compilation there.
<elliottcable> 23 different conferences (!) co-located over 7 days
<elliottcable> unfortunately I've missed a ton of it
<Cheery> I suppose it'd be hard to catch all of it.
<elliottcable> my first time outside my home-country, and my crippling fucking clinical insomnia conspired with jet-lag to leave me incapacitated for about half my time here
<elliottcable> and out of the stuff I've been here for, I've understood maybe 25%, if I'm being generous to myself :x
<elliottcable> but nonetheless there's been some really cool stuff
<elliottcable> actually, everything from IWACO β€” definitely my favourite part so far
<elliottcable> andy wingo's was dense, but I was really proud at myself over how much I actually grasped (nearly all of it, despite not being a heavy Lisp user, nor familiar with CML)
<elliottcable> whereas Kevin Clancy's presentation left me completely fucking lost, b/cuz I'm a dumb
<elliottcable> lessee
<elliottcable> I missed out on this, but wish I hadn't,
<Cheery> there's already some cool stuff in there.
<elliottcable> several *other* works here referenced this work, so it's obv. a big deal,
<Cheery> what's released on public from these later?
<elliottcable> all the proceeds are in the ACM DL
<elliottcable> and ofc many are available preprint from the authors
<Cheery> Is the MLSub guy there?
<elliottcable> if there's preprints available, they're included on the above-linked pages
<elliottcable> also ofc there's my pinboard, if you're not already following it
<elliottcable> MLSub?
<elliottcable> subscribe or w/e
<Cheery> I'll see what I gather from here..
<elliottcable> sometimes papers, sometimes blog posts, sometimes tooling, sometimes formats or whatever
<Cheery> I think there can be something I might be able to use here.
<elliottcable> hopefully can be synthesised into something remotely useful to somebody idk πŸ™„
<elliottcable> bear in mind I've read almost-none of those papers, they're just all on my stack for β€œwhen I get smart enough”
<elliottcable> I've usually read the abstract and some of the background sections to help build my knowledge-base; but any of the actual work is almost universally beyond me
<elliottcable> but yeah β€” the main tracks of these confs are videoed, although idk how they're posted
<elliottcable> you may need to be an ACM member and/or pirate them
<Cheery> well. yeah. I'll think up with my nuggets what can yield and then pursue it like mad, as always.
<elliottcable> Cheery: you'd get along with @small_catgirl
<elliottcable> or nope hm
<elliottcable> @SmallCatgirl
<elliottcable> and ofc @sigfig et al
<Cheery> twitter handles?
<elliottcable> yeah
* elliottcable exhaustedly cuts 2.0 of his dumb boring current library
<Cheery> I think I'm still exploring with Lever.
<Cheery> It's built over the stuff that makes PyPy self-hosting, RPython.
<Cheery> It's got very nice bindings to Vulkan API, and my plans are to do interactive apps with it.
<elliottcable> familiar with rpy, yes
<Cheery> I'm experimenting with a subset of the language that can be compiled on-demand into machine code, so that you would no longer have to write fast parts in C.
<elliottcable> the first talk I linked above did some really cool work with that, showing that some cool optimisations are possible when co-compiling a language runtime with a library, instead of ffi'ing it
<elliottcable> Vulkan?
<Cheery> Successor for OpenGL, though OpenGL will be still supported for a long while.
<elliottcable> oh yeah that
<Cheery> Vulkan is also my bridge to translating subset of Lever to run on a GPU.
<Cheery> that way Lever needs to have the statically typed -part, but it's role is optimization, not "type safety" or "correctness".
<purr\Paws> [giraphe] ELLIOTTCABLE tagged v2.0.0 at 49c468f: https://github.com/ELLIOTTCABLE/giraphe/compare/383a3045e476^...7b499e10bc45
<purr\Paws> giraphe/v2.0.0 bceac94 ELLIOTTCABLE: (merge) Bringing in CI / testing fixes from Current
<purr\Paws> giraphe/v2.0.0 ab31c37 ELLIOTTCABLE: (- fix) Allow partially-applied walking when invoked as a method
<purr\Paws> giraphe/v2.0.0 383a304 ELLIOTTCABLE: (!! new) Support graphs with annotated edges...
<purr\Paws> [giraphe] ELLIOTTCABLE pushed 6 new commits to Current: https://github.com/ELLIOTTCABLE/giraphe/compare/c0c730728845...7b499e10bc45
<purr\Paws> giraphe/Current bceac94 ELLIOTTCABLE: (merge) Bringing in CI / testing fixes from Current
<purr\Paws> giraphe/Current ab31c37 ELLIOTTCABLE: (- fix) Allow partially-applied walking when invoked as a method
<purr\Paws> giraphe/Current 383a304 ELLIOTTCABLE: (!! new) Support graphs with annotated edges...
<elliottcable> y u send two notifications, gh
<elliottcable> stuff gh
<elliottcable> stfu gh
<Cheery> When working on Lever, I think I've noticed few things that can be generally useful and abstract enough to 'take-away'
<Cheery> One idea is that coding in a dynamically typed language gets great when it's algebraic in nature.
<Cheery> certain things in your system means for certain tool for thought.
<Cheery> and when you've found a good one, you use it across the system rather than reimplement it.
<Cheery> for example, C FFI brings the idea of structures, data layouts in the memory.
<Cheery> I didn't roll out a different one for Vulkan buffers to organize data.
<Cheery> I used the one in the C FFI and adjusted it to fit.
<purr\Paws> [giraphe] ELLIOTTCABLE merged Master into Current: https://github.com/ELLIOTTCABLE/giraphe/compare/7b499e10bc45...63ca2d58759b
<Cheery> And that I see is repeating across the board, forming kind of synergies that I didn't thought were possible.
<Cheery> another important idea is the minimal abstraction principle.
<Cheery> It's similar to Chekhov's gun.
<Cheery> you don't create abstractions that you don't use.
<Cheery> and I see that a programming language should support that kind of workflow.
<Cheery> A.K.A. OOP sucks.
<Cheery> then there's the third one. That I don't understand why people have hard time to grok it.
<Cheery> writing a programming language is hard, and you should really do decisions that help you develop the language faster.
<Cheery> all of them.
<Cheery> because you don't really know whether it's good until you've had it in front of you for weeks.
<Cheery> even if it took 20 minutes to compile the runtime, and it spent 10 seconds at parsing the libraries into bytecode.
<Cheery> if those are coming off from tradeoffs that sped up the development of the language, they've been good decisions.
<Cheery> all right. that's all.
<Cheery> I know your reading list must be dying, but here's one more: http://tratt.net/laurie/blog/entries/what_challenges_and_trade_offs_do_optimising_compilers_face.html
<Cheery> elliottcable: thanks for the links.. there ARE some interesting things in there.
_whitelogger has joined #elliottcable