| --- Log opened Fri Dec 16 00:00:37 2016 |
| 00:34 | | catadroid [catalyst@Nightstar-54p.06s.132.82.IP] has joined #code |
| 00:37 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has quit [[NS] Quit: Leaving] |
| 01:03 | | Vorntastic [Vorn@Nightstar-quk3ol.sub-174-199-3.myvzw.com] has quit [[NS] Quit: Bye] |
| 01:39 | | Vash [Vash@Nightstar-uhn82m.ct.comcast.net] has joined #code |
| 01:51 | | Derakon_ is now known as Derakon |
| 01:51 | | mode/#code [+ao Derakon Derakon] by ChanServ |
| 02:28 | | catadroid` [catalyst@Nightstar-uu6cgc.dab.02.net] has joined #code |
| 02:30 | | CoolDude [NSwebIRC@Nightstar-l8t4dn.mi.comcast.net] has joined #code |
| 02:32 | | catadroid [catalyst@Nightstar-54p.06s.132.82.IP] has quit [Ping timeout: 121 seconds] |
| 02:35 | < CoolDude> | hii |
| 02:36 | | CoolDude [NSwebIRC@Nightstar-l8t4dn.mi.comcast.net] has left #code ["IRC.Brett-TechRepair.com 6667 - CoolDudeChat.ZohoSites.com"] |
| 03:00 | <@Alek> | ..... |
| 04:40 | | celticminstrel [celticminst@Nightstar-h4m24u.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 04:50 | <~Vornicus> | I am finally learning to use home and end to navigate lines. |
| 04:51 | <~Vornicus> | It's weird how that tool never came up, so I need to consciously train myself to use it |
| 04:58 | | * McMartin is full of cake, opens up advent of code for the countdown |
| 05:00 | | * Vornicus got around to doing 9B. Hooray recursion! Could probably go back and replace 9A with a very similar implementation with one fewer recurse call. |
| 05:04 | | Derakon is now known as Derakon[AFK] |
| 05:25 | | Vash [Vash@Nightstar-uhn82m.ct.comcast.net] has quit [Connection closed] |
| 05:30 | <~Vornicus> | ok 10 looks fun. |
| 05:52 | | himi [sjjf@Nightstar-dm0.2ni.203.150.IP] has quit [Ping timeout: 121 seconds] |
| 07:19 | | Kindamoody[zZz] is now known as Kindamoody |
| 07:22 | | Kindamoody is now known as Kindamoody|afk |
| 07:25 | | catadroid` is now known as catadroid |
| 07:26 | < catadroid> | I'm very happy at how my ideas are coming together |
| 07:28 | <&McMartin> | :D |
| 07:28 | < catadroid> | I've never been this sustained with something not work related |
| 07:28 | <&McMartin> | woot |
| 07:29 | | * McMartin meanwhile has properly buried an abandoned project, brought another to release candidate status, and found the ancillary materials needed to finalize the release on two more. |
| 07:29 | <&McMartin> | It's the end-of-year cleanup/mercykill, I guess |
| 07:30 | <&McMartin> | I've got two projects I'd like to work on in earnest, but I will not let myself start until these old ones are cleaned up and sent out. |
| 07:30 | <&McMartin> | (They a work of art is never finished, only abandoned. The one I just got to release candidate status is less "abandoned" and more "uploaded to an archive in the dead of night, followed by fleeing cackling into the wilderness") |
| 07:32 | < catadroid> | Nice |
| 07:37 | < catadroid> | I'm still debating the best way to evaluate my language |
| 07:37 | < catadroid> | It feels like something stack based would be reasonable |
| 07:38 | < catadroid> | At least as a first attempt |
| 07:41 | < catadroid> | I've been really enjoying the general mode of thought that Reduct promotes |
| 07:45 | <~Vornicus> | oh man strong typing just stung me. |
| 07:45 | | * Vornicus is way too used to js apparently |
| 07:46 | <&McMartin> | Watcha workin' on? |
| 07:46 | <&McMartin> | in |
| 07:46 | <&McMartin> | but also on |
| 07:47 | <~Vornicus> | python, AoC 10 |
| 07:48 | <~Vornicus> | '7' > '23' |
| 07:48 | <&McMartin> | Ah yes |
| 07:50 | < catadroid> | :) |
| 07:51 | < catadroid> | It annoys me that I'm reasonably competent with python |
| 07:51 | < catadroid> | It feels like such a nothing language |
| 07:51 | <&McMartin> | It's my shell script |
| 07:51 | <&McMartin> | Or given my sordid and twisted past, it's really my QBASIC |
| 07:52 | < catadroid> | But Ruby and Perl are so much better for that, at least IME |
| 07:52 | < catadroid> | my QBASIC is C++, which probably explains a lot |
| 07:52 | | * Vornicus typically thinks in Excel and then either JS or Python |
| 07:52 | <&McMartin> | I hold an unreasoning hatred for Ruby and while I've dealt with Perl in some capacity I prefer not to. |
| 07:52 | <~Vornicus> | I have a *perfectly reawsonable* hatred for Ruby |
| 07:53 | <~Vornicus> | BUt that's a story I'm pretty sure you've all heard |
| 07:53 | <&McMartin> | ???!!??:!!!?? |
| 07:53 | <~Vornicus> | That is not it. |
| 07:53 | < catadroid> | Huh? Nope! Do tell |
| 07:53 | <~Vornicus> | oh, you'll like this! It's a type problem |
| 07:54 | <~Vornicus> | So, I was planning on building vectors (of the mathematical type) |
| 07:54 | <&jerith> | My best debugging story involves Ruby. But I'll wait for Vornicus to finish. :-) |
| 07:54 | <~Vornicus> | One thing you can do with vectors is you can multiply or divide them by regular numbers; x*v, v*x, and v/x all work. ...but x/v makes no sense. |
| 07:58 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has joined #code |
| 07:58 | | mode/#code [+o himi] by ChanServ |
| 07:58 | <~Vornicus> | Various languages that handle operator overloads handle this different ways: in C++ you make two of the operator methods on Vector and the third one as a bare function that's a friend; in Python you give Vector __mul__, __rmul__, and __div__; in Lua Vector's __mul and __div will get their two arguments in order so you just reject the reversed version on the latter. |
| 07:58 | <~Vornicus> | Ruby ... does none of these things. |
| 08:00 | <~Vornicus> | In Ruby, operators are a two step process: 1. a coercion that possibly converts the two operands, and 2. a single call to the first operand's operator overload function |
| 08:04 | <~Vornicus> | So here's the puzzle: how do you manage to make it so x*v, v*x, and v/x work, but x/v, x+v, x-v, v+x, v-x, v*v, and v/v don't? Coercion is a problem: you can't really pretend to expand the scalar into the vecotr and then use componentwise product etc, because all those other things *shouldn't* work and now *do* |
| 08:05 | < catadroid> | Yeah... that sounds unhelpful |
| 08:05 | <~Vornicus> | But if you *don't* coerce, you have *another* problem: you have to duplicate your effort for every single numeric type. While there is a numeric abstract base class, its operator functions never get called, having been (obviously) overloaded by its implementors like int and real |
| 08:07 | <~Vornicus> | So -- and this is in keeping with Ruby's philosophy! -- you go in and you move aside the official implementations, implement your own that knows about this new vector class, and for things that aren't the vecotr class you call the original. ...except of course that you have to do this half a dozen times, once for each already-existing numeric type. |
| 08:08 | <~Vornicus> | I explained this problem to a friend who was a huge ruby evangelist and he sat there for three hours trying to puzzle out a way where I didn't have to repeat myself so much and was just "...shit, you're stuck" |
| 08:10 | < catadroid> | That sounds like when u attempted to write performant statistical code in java |
| 08:10 | < catadroid> | When I* |
| 08:10 | | * Vornicus also eyes AoC 11. Madness! |
| 08:10 | | * Vornicus gives jerith the rubber chicken |
| 08:11 | <&jerith> | Vornicus: That's the only one where I'm not reasonably sure my solution is general. |
| 08:11 | <&jerith> | So, my debugging story. |
| 08:11 | <&jerith> | We had some code that had been working perfectly in production for months. |
| 08:12 | <&jerith> | While adding a new feature, we discovered a tyop in a well-used code path: "false" had been spelled "flase". |
| 08:12 | <&jerith> | This should be, at minimum, a "missing identifier" error. |
| 08:12 | <~Vornicus> | Um |
| 08:13 | <~Vornicus> | Holy Shit, Fuck Ruby |
| 08:14 | <&jerith> | We searched the entire codebase and all its dependencies for "flase", found one other example... a similar tyop in an explanatory comment. |
| 08:14 | <&jerith> | Our first clue was when we logged the value of "flase" at that point in the code: nil |
| 08:15 | <~Vornicus> | I forget if ruby does some crazybullshit where nil and undefined are the same thing |
| 08:16 | <&jerith> | Eventually, after half a day of digging around and trying various things, we discovered that some class somewhere had method_missing defined. |
| 08:16 | <&McMartin> | Oh wait |
| 08:16 | <&McMartin> | This is that thing that's listed under Let's Talk About Ruby |
| 08:16 | <&jerith> | This was somehow being indirectly included in the class we were looking at. |
| 08:17 | <~Vornicus> | "the wat is that someone would ever do this" |
| 08:17 | <&jerith> | Now, the *correct* way to use method_missing is to check if the name and args are sensible and throw an exception if they aren't. |
| 08:18 | <&jerith> | In this case, in the limited context in which this was *supposed* to be used, returning nil wasn't the worst behaviour ever. |
| 08:19 | <&McMartin> | "This is actually a result of how awesome Ruby is, but if you actually *do* this, then wat" |
| 08:19 | <&jerith> | Except that because of various things happening over time, that method_missing ended up escaping from its original confines. |
| 08:19 | | * Vornicus got the quote wrong |
| 08:20 | <&jerith> | There's nothing wrong with method_missing conceptually. |
| 08:20 | <&jerith> | Python's __getattr__ is basically the same idea with much cleaner semantics. |
| 08:21 | <&McMartin> | True, though even there, it is well-attributed that calling __hasattr__ three times will cause the King in Yellow to arrive and devour your sanity |
| 08:21 | <&jerith> | (Python's semantics can be much cleaner because method calls *must* be applied to a specific object.) |
| 08:22 | <&jerith> | McMartin: Again, there are a few places where This Is The Correct Solution. |
| 08:22 | <&McMartin> | Yep |
| 08:22 | <&McMartin> | My only major published work in Python is one of those places |
| 08:22 | <&McMartin> | Python's idiomatic equivalent to Visitor |
| 08:23 | <&jerith> | The thing I found really annoying about Ruby's metaprogramming is that it's really hard to limit scope. |
| 08:24 | <&jerith> | You can trivially add a method to someone else's class, but it's really hard to add a method to a single object. |
| 08:25 | <~Vornicus> | where "someone else" includes core objects such as string. |
| 08:25 | <&jerith> | Yes. |
| 08:26 | <&McMartin> | Adding a method to a single object implies things about your object ontology that I do not consider default |
| 08:27 | <&jerith> | It's a thing I use in tests much more than real code. |
| 08:27 | <&jerith> | Patch a method on an object to behave differently for this particular test case, etc. |
| 08:28 | <&jerith> | I can't remember the specific reason I wanted it in Ruby. |
| 08:29 | <&jerith> | It probably had something to do with getting an object I could mutate but not replace and needing to add behaviour to it. |
| 08:30 | <&jerith> | And not wanting that to infect all the other things that were using this object, because they might also need to modify it in similar ways. |
| 08:33 | <&jerith> | Err, all the other things using objects of this type. |
| 08:47 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 08:58 | | catadroid` [catalyst@Nightstar-gtbck2.dab.02.net] has joined #code |
| 09:00 | | catadroid [catalyst@Nightstar-uu6cgc.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 10:23 | | Syloq [Syloq@NetworkAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 11:31 | | catadroid` is now known as catadroid |
| 11:31 | < catadroid> | Fun things - people telling me my code is awesome and easy to maintain |
| 11:31 | | * catadroid sits and looks content with life |
| 11:36 | <@TheWatcher> | \o/ |
| 11:50 | <&jerith> | catadroid: :-D :-D :-D |
| 12:22 | <&jerith> | Yay, AoC 16 done. |
| 12:23 | <&jerith> | I knew my naive solution for part 1 wasn't going to be performant enough for part 2, but I ran it anyway while I built a real solution. |
| 12:23 | <&jerith> | Four and a half hours later it hadn't finished. |
| 13:26 | <@TheWatcher> | Good gods |
| 13:26 | <@TheWatcher> | It's far harder to get a fucking input box to accept voice input than it should be |
| 13:34 | | celticminstrel [celticminst@Nightstar-h4m24u.dsl.bell.ca] has joined #code |
| 13:34 | | mode/#code [+o celticminstrel] by ChanServ |
| 14:15 | <&jerith> | Voice input? |
| 14:16 | <&jerith> | Oh, I see. |
| 14:16 | <&jerith> | Why is this not a thing that all inputs get by default? |
| 14:17 | <@TheWatcher> | This is an excellent question! |
| 14:17 | <@TheWatcher> | One I have no answer for other than "browser developers are fucking trolls" |
| 14:19 | <@TheWatcher> | For a while there, chrome supported a x-webkit-speech attribute that marked input boxes as supporting speech input, added a little mic icon to the side, click on it, speak, and voila. |
| 14:23 | <@TheWatcher> | But now, that's defunct; everyone's supposed to use the speech API |
| 14:24 | <@TheWatcher> | So you need to write fuckign client-side javascript code to do something that should be part of the standard browser-provided widgets |
| 14:30 | <@TheWatcher> | (but then, I shouldn't really be surprised; this sort of bullshit is par for the course with web dev) |
| 15:18 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 15:18 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 15:25 | | * Vornicus examines 11 with fresh eyes, determines that it's not that bad at all |
| 15:28 | | catadroid` [catalyst@Nightstar-2c36nq.dab.02.net] has joined #code |
| 15:30 | <~Vornicus> | the validity and pathing rules are a little weird, but that's all. |
| 15:30 | | catadroid [catalyst@Nightstar-gtbck2.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 16:00 | <&ToxicFrog> | I should do 8 today |
| 16:02 | < catadroid`> | I'd start with one and see how that goes |
| 16:02 | < catadroid`> | Eight in a day sounds like a lot of work |
| 16:07 | | catadroid` is now known as catadroid |
| 16:15 | <&ToxicFrog> | Har har :P |
| 17:29 | | * Vornicus writes the state transition system. Ewk. |
| 17:31 | <~Vornicus> | One of the big difficulties with pathfinding complicated stuff like this is that often you've got a thing where the natural method of defining new states is a mutation, but you need immutable objects for state to work. |
| 17:32 | | * Vornicus makes mutify and immutify |
| 17:32 | | Syloq [Syloq@NetworkAdministrator.Nightstar.Net] has joined #code |
| 17:32 | | mode/#code [+o Syloq] by ChanServ |
| 17:41 | <@celticminstrel> | ...what... |
| 17:44 | <~Vornicus> | ? |
| 17:44 | <@celticminstrel> | Mutify and immutify |
| 17:46 | <~Vornicus> | They're highly specific to my actual task here; they take the immutable object that I've made state things out of and return a thing that's mutable -- or vice versa |
| 17:48 | <~Vornicus> | they don't actually turn an immutable object actually mutable |
| 17:49 | <@celticminstrel> | I see. |
| 17:52 | < catadroid> | Freeze and thaw |
| 17:53 | <&jerith> | I like Vorn's names better. :-) |
| 17:54 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has joined #code |
| 17:55 | < catadroid> | Break and fix |
| 17:55 | < catadroid> | Unborrow and borrow |
| 17:56 | <&jerith> | turn_into_horrible_dangerous_thing and create_nice_clean_safe_thing :troll: |
| 18:41 | <~Vornicus> | oy there are a few possible states |
| 18:42 | <~Vornicus> | (duh. ten objects, four floors, a first pass suggests 2^22 possible states, though many are not allowed) |
| 18:48 | | * Vornicus accidentally blows up the universe. |
| 18:52 | <&McMartin> | I need to have a go at doing the representation change someone suggested to me after the fact |
| 18:52 | <&McMartin> | That one would have, I think, made it feasible to not need mutify/immutify either. |
| 18:53 | <~Vornicus> | I can envision a method to encode as integers... |
| 18:55 | <&McMartin> | This was a representation change that altered big chunks of the algorithm. It wasn't a mere encoding shift from what I was doing. |
| 18:55 | <&jerith> | All my state was immutable, but I'm using a language that makes that very natural. |
| 18:55 | <&jerith> | I'm also up to day 7 of last year's. |
| 18:56 | <~Vornicus> | ah. I wrote a* as generic |
| 18:56 | <&McMartin> | I was using straight BFS |
| 18:56 | <&McMartin> | That part doesn't change |
| 18:56 | <&McMartin> | The rest of it does |
| 18:57 | | * Vornicus destroys the universe again without having promiscuous diagnostics |
| 18:57 | <~Vornicus> | *definitely* need a smaller representation. |
| 18:58 | <&jeroud> | I used a greedy DFS. |
| 18:59 | <&jeroud> | It works very nicely on all the input I've thrown at it, which isn't very much. |
| 19:00 | <&jeroud> | Hrm. Maybe I should borrow some input from you guys for it. |
| 19:00 | <~Vornicus> | wait how do you get shortest path from dfs |
| 19:07 | <&jeroud> | By greedy. |
| 19:15 | <&jeroud> | I always move the lowest things first. |
| 19:16 | <&jeroud> | I always move two things up unless there's only one thing on that floor |
| 19:16 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection reset by peer] |
| 19:17 | <&jeroud> | Oh, Vorn disappeared. |
| 19:24 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 19:24 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 19:26 | | * Vornicus fiddles, figures out how to manipulate state with just arithmetic, works on state legality. |
| 20:21 | | ion [Owner@Nightstar-6grqph.vs.shawcable.net] has quit [Ping timeout: 121 seconds] |
| 20:44 | | Vash [Vash@Nightstar-uhn82m.ct.comcast.net] has joined #code |
| 21:20 | | ion [Owner@Nightstar-6grqph.vs.shawcable.net] has joined #code |
| 21:20 | | mode/#code [+o ion] by ChanServ |
| 21:32 | <~Vornicus> | there we are. takes 2m29s, but it does not explode my memory. |
| 21:35 | <~Vornicus> | oy. another four things would probably be bad. |
| 21:41 | | catalyst [catalyst@Nightstar-bt5k4h.81.in-addr.arpa] has quit [[NS] Quit: Leaving] |
| 21:45 | <~Vornicus> | ...would help if my a* heuristic actually worked. |
| 21:46 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 21:58 | | catadroid` [catalyst@Nightstar-3voi6p.dab.02.net] has joined #code |
| 22:01 | | catadroid [catalyst@Nightstar-2c36nq.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 22:16 | <&McMartin> | Oh dear. An Underhanded Rust contest has been announced. https://underhanded.rs/ |
| 22:55 | | Kindamoody|afk is now known as Kindamoody |
| 23:09 | <&ToxicFrog> | Awesome. |
| 23:12 | | * Vornicus has several a* fails, wonders what the heck he's doing wrong. |
| 23:12 | <@abudhabi> | Aaaaaaaaaaaa*! AAaaaaaaaa*! |
| 23:21 | | Kindamoody is now known as Kindamoody[zZz] |
| --- Log closed Sat Dec 17 00:00:38 2016 |