| --- Log opened Sat Feb 13 00:00:16 2016 |
| 00:05 | | kylo-ren [cmnd@Nightstar-nelvj0.dyn.optonline.net] has quit [Connection closed] |
| 00:44 | | Derakon[AFK] is now known as Derakon |
| 01:53 | | Derakon is now known as Derakon[AFK] |
| 03:50 | | celticminstrel [celticminst@Nightstar-3nv7ni.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 04:31 | | macdjord|slep [macdjord@Nightstar-r9vt2h.mc.videotron.ca] has quit [Ping timeout: 121 seconds] |
| 05:04 | | Crossfire [Z@Nightstar-r9lk5l.cust.comxnet.dk] has quit [Ping timeout: 121 seconds] |
| 05:07 | | * Vornicus gnaws his brain, tries to get through the basic engineering so he can get to the part where he actually does the thing he wanted to actually do |
| 05:24 | | * Vornicus determines he does not really know what the hell kind of object inheritance structure he wants and he doesn't know how to make any of them from scratch anyway. |
| 05:27 | <&McMartin> | Is this for Vorns ahoy and why are you reinventing not merely the wheel but the manual transmission |
| 05:29 | <~Vornicus> | Because I'm a dumbass |
| 05:30 | <&McMartin> | It turns out I have Opinions about Polymorphism |
| 05:30 | <~Vornicus> | Part of this is, I'm working in Lua, and fuckin' nothing is just there in Lua. |
| 05:30 | <&McMartin> | If you have an intended implementation language I can probably talk about... aha. |
| 05:30 | <&McMartin> | ... is there a reason you're using Love2d instead of Pygame here |
| 05:31 | <~Vornicus> | Because every time I've tried pygame I had to write my own damn *event loop* |
| 05:31 | <&McMartin> | Okay |
| 05:31 | <~Vornicus> | And if I don't, or if I screw it up in any of the subtle ways that it's possible to screw that up, my program hangs. |
| 05:31 | <~Vornicus> | And that part at least I don't have to do in love2d |
| 05:32 | <&McMartin> | I would start with tables that map constants to callable objects |
| 05:32 | <&McMartin> | No inheritance beyond duck typing becuase fuck inheritance, and it's closer to Python and allows monkey-patching |
| 05:32 | <&McMartin> | Also obviously Vorns Ahoy's semantics should involve duck typing. |
| 05:33 | <&McMartin> | THE COSMOS COMMANDS IT |
| 05:33 | <&McMartin> | The All-Father is fond of dad jokes |
| 05:33 | <&McMartin> | Once inheritance becomes necessary I would seriously consider prototypal inheritance, but I forget how to actually do that in Lua. |
| 05:34 | <~Vornicus> | But then with lua i'm like -- okay how do I want this to work and I have to start at the bottom and I don't know how to name things and then I'm like -- shit, this interface is the same all over the place. Everything has an update and it does a lot of the same things and a little of the different things |
| 05:34 | <&McMartin> | Duck typing at the actual lua level, I think, and then track protocols by hand |
| 05:35 | <~Vornicus> | So if I then make congo bongo or donkey kong I do monkey patching |
| 05:35 | <&McMartin> | Makes sense to me |
| 05:36 | <~Vornicus> | (man, fuckin' congo bongo) |
| 05:36 | <&McMartin> | I think the way I'd do this is not have true inheritance, but have template objects for each class |
| 05:36 | <&McMartin> | Instantiating the class is cloning that table |
| 05:36 | <&McMartin> | And it preloads the non-monkeypatched methods &c |
| 05:37 | <&McMartin> | "Inheritance" can be worked in by hand by having some templates share implementations of some methods |
| 05:38 | <&McMartin> | Lua's method call syntax works well with this IIRC |
| 05:40 | <~Vornicus> | Yeah, lua has pretty convenient stuff for this -- I made a set class one time and I forget what it was for but it was not hard. |
| 05:41 | <&McMartin> | If you want Actual vtables it's more of a bummer |
| 05:41 | <&McMartin> | but you don't want those, because Quack |
| 05:45 | <~Vornicus> | Ok so: 1. make any methods I want to be common as just functions and then assign them in where I want them, or in the most complex case make a table with them all in it and then clone + patch for polymorph. 2. Other than that, things do whatever the hell they want, so long as they follow the necessary interface/protocol, which is explicit in my docs. 3. quack. |
| 05:45 | <&McMartin> | quack. |
| 05:46 | <&McMartin> | 2 may just be the most convenient way to lay them out even without patching |
| 05:48 | <~Vornicus> | This works out because (for instance) I have things that need to draw both before and after things it contains (ducks can be indoors or out, and the gondola can go behind tall roofs), so not enforcing a particular method for draw or update works better. |
| 05:49 | <&McMartin> | A good thing to do with that also is to sort drawables in priority order and draw in reverse priority order |
| 05:52 | <~Vornicus> | It is true. |
| 05:52 | <~Vornicus> | Perhaps "just declare your drawables and return them, don't actually draw anything yourself" is a decent idea |
| 05:57 | <&McMartin> | Now I am wondering what all lives at the code zoo |
| 05:57 | <&McMartin> | duck types, monkey patching, yak shaving |
| 05:57 | <~Vornicus> | I can't name a game with yaks so that's not happening |
| 05:57 | <~Vornicus> | though I did quite a bit of yak shaving lately as you can tell |
| 05:58 | <&McMartin> | this is unrelated to games now, just coding terms that are animal-related |
| 06:08 | <~Vornicus> | two kinds of ducks |
| 06:08 | <~Vornicus> | one's rubber |
| 06:14 | <&McMartin> | Ah yes |
| 06:14 | <&McMartin> | Also goldfish for that one |
| 06:14 | <&McMartin> | Also augh, I am tired |
| 06:15 | <&McMartin> | And yet these cookies have not been eaten yet |
| 06:16 | | macdjord [macdjord@Nightstar-r9vt2h.mc.videotron.ca] has joined #code |
| 06:16 | | mode/#code [+o macdjord] by ChanServ |
| 06:29 | | macdjord [macdjord@Nightstar-r9vt2h.mc.videotron.ca] has quit [[NS] Quit: I have been... and always shall be... your friend.] |
| 06:42 | | Derakon[AFK] is now known as Derakon |
| 06:58 | | Derakon is now known as Derakon[AFK] |
| 07:20 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection closed] |
| 07:23 | <&jerith> | I find that dog debugging works better than rubber duck debugging. |
| 07:23 | <&jerith> | Dogs listen more attentively. |
| 07:23 | <&jerith> | And are happy to go away if your problem is too boring. |
| 08:14 | | Kindamoody[zZz] is now known as Kindamoody |
| 08:28 | | macdjord [macdjord@Nightstar-r9vt2h.mc.videotron.ca] has joined #code |
| 08:28 | | mode/#code [+o macdjord] by ChanServ |
| 09:01 | | catadroid [catalyst@Nightstar-7ffc9s.cable.virginm.net] has joined #code |
| 09:13 | | gizmore [kvirc@Nightstar-ticgdn.dip0.t-ipconnect.de] has joined #code |
| 09:31 | | gizmore|2 [kvirc@Nightstar-ticgdn.dip0.t-ipconnect.de] has joined #code |
| 09:34 | | gizmore [kvirc@Nightstar-ticgdn.dip0.t-ipconnect.de] has quit [Ping timeout: 121 seconds] |
| 09:47 | | gizmore|2 [kvirc@Nightstar-ticgdn.dip0.t-ipconnect.de] has quit [[NS] Quit: KVIrc 4.9.1 Aria http://www.kvirc.net/] |
| 10:18 | | ErikMesoy [Erik@Nightstar-hq72t5.customer.cdi.no] has quit [Ping timeout: 121 seconds] |
| 11:01 | | ErikMesoy [Erik@Nightstar-hq72t5.customer.cdi.no] has joined #code |
| 11:38 | | catadroid [catalyst@Nightstar-7ffc9s.cable.virginm.net] has quit [[NS] Quit: Bye] |
| 12:43 | | Kindamoody is now known as Kindamoody|out |
| 13:19 | <@abudhabi> | https://boingboing.net/2016/02/12/nigerian-astronaut-lost-in-spa.html |
| 13:21 | <@gnolam> | Hee |
| 14:29 | <@abudhabi> | It's nice to see that the USB hub I have works even unpowered. |
| 14:54 | | celticminstrel [celticminst@Nightstar-3nv7ni.dsl.bell.ca] has joined #code |
| 14:54 | | mode/#code [+o celticminstrel] by ChanServ |
| 16:03 | | Crossfire [Z@Nightstar-r9lk5l.cust.comxnet.dk] has joined #code |
| 16:03 | | mode/#code [+o Crossfire] by ChanServ |
| 16:30 | | gnolam [lenin@Nightstar-t1tbf0.cust.bahnhof.se] has quit [[NS] Quit: The speech pleases him. Lord Ipsulot will be spared, and the worlds of Bok will not be smashed.] |
| 18:08 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 18:08 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 18:19 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has joined #code |
| 18:29 | <~Vornicus> | ...huh. It occurs to me as I write this how rarely I use a for loop without a collection to iterate over |
| 18:29 | < gizmore> | java sucks |
| 18:30 | <~Vornicus> | you say this as though I've used java in the past, uh, 12 years |
| 18:31 | <@abudhabi> | Java is OK. |
| 18:32 | <~Vornicus> | though nwo it occurs to me you might not have been replying to me and I apologize for the misunderstanding |
| 18:53 | <&jerith> | That was either a reply to you or a very strange non-sequitur. |
| 19:00 | <@abudhabi> | It is quite a strange reply too! |
| 19:01 | < ErikMesoy> | My memories of java are an endlessly repeating chant of "public static void main string args". |
| 19:06 | | Kindamoody|out is now known as Kindamoody |
| 19:06 | <&jerith> | Java is not a very nice language. |
| 19:06 | <&jerith> | But it has its uses. |
| 19:26 | <&McMartin> | You should only be saying that once. |
| 19:33 | | Derakon[AFK] is now known as Derakon |
| 19:33 | <&Derakon> | Erik: never did any GUI work in Java, eh? foo.addActionListener(new ActionListener() {@Override public void actionPerformed(ActionEvent e) {actually do something;}}); |
| 19:37 | <&McMartin> | Derakon: Right, but people don't usually object to that the way they appear to object to classes-as-mandatory-namespace or to access modifiers |
| 19:37 | <&McMartin> | Er, or to nonmodal access modifiers, rather |
| 19:38 | <&Derakon> | "Access modifiers"? |
| 19:38 | <~Vornicus> | public, private, protected |
| 19:38 | <&Derakon> | Classes-as-mandatory-namespaces is moderately annoying, but I can cope with it. Java's infinite verbosity and awful support for reflection and lambdas (we're still stuck with Java 6 :( ) are the things that annoy me most. |
| 19:38 | <&Derakon> | Ah. |
| 19:39 | <~Vornicus> | specifically, they complain that in C++ you can go public: foo... bar... baz... private: quux... blarg... floogy... but in Java you have to mention each one. |
| 19:39 | <@abudhabi> | What really annoys me about Java is that it has null. |
| 19:39 | <@abudhabi> | And NPE. |
| 19:40 | <&Derakon> | The only thing about Java's null that bugs me is that I can't compare (potentially null variable A) to (potentially null variable B). |
| 19:41 | <&Derakon> | Whereas e.g. Python will let you compare (potentially None variable A) to (potentially None variable B) and return true if they're both None or (neither is None and they compare equally in whatever type they have) |
| 19:41 | <@abudhabi> | I would like a pirate elvis operator. |
| 19:41 | <@abudhabi> | And optionality that isn't horrible. |
| 19:43 | <&McMartin> | Don't settle for optionality |
| 19:43 | <&McMartin> | Get proper variants |
| 19:43 | <&McMartin> | As usual, "proper" for both means "like Haskell does it" >_> |
| 19:43 | <&Derakon> | Optionality, as in optional function arguments? That'd be nice to have. |
| 19:43 | <&Derakon> | As would be named parameters. |
| 19:43 | <&McMartin> | I was reading it as nullable types |
| 19:43 | <@abudhabi> | Yes. |
| 19:43 | <&Derakon> | Ah, yes. |
| 19:43 | <@abudhabi> | Java already das optional params, sort of. |
| 19:43 | <&McMartin> | e.g, variables that are "thing that is type X, or None, and nothing else" |
| 19:44 | <@abudhabi> | There's ... and all. |
| 19:44 | <&Derakon> | That doesn't count. |
| 19:44 | <&McMartin> | ... is variadic, not optional |
| 19:44 | <@abudhabi> | It can be 0 or more! That sounds optional to me. |
| 19:44 | <&McMartin> | Java does optional arguments by defining each combination and relying on overloads for the rest. |
| 19:45 | <&McMartin> | ... is an infinite number of those, so it requires a separate feature. |
| 19:45 | <&Derakon> | I mean, you could conceivably do optional with variadic by taking an Object... and then examining the number of parameters passed in and selecting defaults as needed. |
| 19:45 | <&Derakon> | But holy crap that would be awful. |
| 19:45 | <&McMartin> | ... No, that's not how it's done |
| 19:45 | <&McMartin> | If you have three arguments, the last two of which are optional |
| 19:45 | <&McMartin> | You define three functions with the same name |
| 19:45 | <&Derakon> | Yes, I know. |
| 19:45 | <&McMartin> | Two of which are one-line forwards |
| 19:45 | <&Derakon> | That is also awful. |
| 19:45 | <&McMartin> | This is actually, arguably, better than C++ optional arguments |
| 19:45 | <&McMartin> | Where what the optional argument is depends on which header you included. |
| 19:46 | <&McMartin> | The correct thing is to take something that looks like the second and compile it to the first~ |
| 19:47 | <&McMartin> | So that changing the default value doesn't necessitate recompiling all code that links against it |
| 19:47 | <&McMartin> | Unless you're in the camp that says this is a feature~ |
| 19:48 | <&McMartin> | That said, what you described as "holy crap that would be awful" is a thing Clojure can do trivially |
| 19:48 | <&McMartin> | Because it turns out destructuring pattern-match as a basic control flow construct is hax~ |
| 19:50 | <&Derakon> | I meant it would be awful in Java. |
| 19:50 | <&Derakon> | Just imagining the "okay, I have 3 arguments, which means I cast the first to an Integer, the second to a Rectangle, etc. etc. etc." |
| 19:51 | <&McMartin> | I bet you could get that into a pretty idiomatic form |
| 19:51 | <&McMartin> | something like a chain of Integer i = args.length >= 1 ? args[0] : new Integer(3); |
| 19:52 | <&McMartin> | I guess there'd be the cast |
| 19:52 | <&McMartin> | And of course you've lost compile-type checking of arguments there. |
| 19:53 | <&Derakon> | Yep. |
| 19:53 | <&Derakon> | Would it stop you from passing an int or double or the like in? |
| 19:53 | <@abudhabi> | Specifically what I would want: some.chain.of.delegation() to return null instead of throwing NPE if it tries to dereference one of those nulls, and transcribing optional variables as empty strings instead of "null" or throwing NPE because null.toString() doesn't exist. |
| 19:57 | <~Vornicus> | you don't get that former in, like, any sensible language |
| 19:58 | <@abudhabi> | What about the languages that implement '.?' in addition to '.'? |
| 19:59 | <&ToxicFrog> | Vornicus: I know that I've worked with a few where "nil.foo" and "nil.foo()" are both nil, i.e. they behave similar to NaN, but I can't remember names offhand |
| 19:59 | <&ToxicFrog> | (you can get lua to behave like this with __index and __newindex, but you probably shouldn't) |
| 20:02 | <&McMartin> | Objective-C allows you to invoke any method in the universe on nil, and it returns nil |
| 20:02 | <&McMartin> | However |
| 20:02 | <&McMartin> | This is only a feature if you believe that a program should never halt execution |
| 20:03 | <&McMartin> | fail-fast vs. fail-soft is a design decision |
| 20:07 | <@abudhabi> | I'm a web dev. If the website crashes, that's awful. |
| 20:07 | <@abudhabi> | If the user gets a blank page due to NPE instead of a formal error message, that's not good either. |
| 20:13 | <~Vornicus> | This does not seem like the kind of thing that would be fixed by allowing nil pollution |
| 20:13 | <&McMartin> | And if a user gets "nil" instead of the content you intended becaues the value propagated from any of the past 15,000 operations, that's just as bad but there's no stack trace because the language says that's not an error. |
| 20:14 | <&McMartin> | But yes, that is the entire argument. |
| 20:14 | <&McMartin> | Fail-soft says "programs aren't supposed to stop working" |
| 20:14 | <&McMartin> | Fail-fast says "when they do, they should stop at the point where the problem actually happened, not when life support finally gave out" |
| 20:14 | <&ToxicFrog> | My general use case for that is "I have a chain of operations, some of which may return nil, and I want to nil check once at the end of the chain rather than checking every single operation, because the fail case is the same regardless" |
| 20:14 | <&McMartin> | Right |
| 20:15 | <&McMartin> | This is where I say "you actually want a variant there", and this is where Rust's try! macro carries the day |
| 20:15 | <&McMartin> | Though I'm not 100% sure that works cleanly for operations whose error types are different error types. |
| 20:19 | <&McMartin> | Ah, no, try! isn't the one I'm thinking of. |
| 20:19 | <&ToxicFrog> | Tragically, I find that I usually run into this in languages that don't have such things. |
| 20:19 | <&ToxicFrog> | Actually, thinking back on it, usually Java. |
| 20:19 | <&McMartin> | Well |
| 20:20 | <&McMartin> | That's because the only languages with variant types are the ML-descendants, so you've got, um, ML, Haskell, and Rust, basically. |
| 20:20 | <&McMartin> | Maybe swift, but they keep mutating how those work to fit with the ObjC ABI and I'm not sure they do an equivalent of Result. |
| 20:22 | <&McMartin> | But yeah, C#'s .? is basically a nilcheck at the end for the sequence you describe, and it also has i ?? x which is i != null ? i : x |
| 20:23 | <&ToxicFrog> | Clojure has some-> and some->>, which is basically exactly what I want |
| 20:23 | <@abudhabi> | Which would be great in my use cases, where I usually remember to check for getting a null value out of the chain, but not that each particular element of the chain is not null itself. |
| 20:24 | <&ToxicFrog> | (-> v f g h) is eqv to (h (g (f v))); ->> is similar but has different behaviour if the functions already have arguments. the some- variants short out to nil if any operation in the chain returns nil. |
| 20:26 | <&McMartin> | Yeah. That particular operator is useless if you aren't primarily functional, though |
| 20:26 | <&McMartin> | The member access equivalent in C# is ?g.?h.?f(v) |
| 20:26 | <&McMartin> | er, minus the first ? |
| 20:27 | <&McMartin> | Also IIRC, it has been quite a while since I have used an Ada-class language |
| 20:30 | <&McMartin> | (Seriously, beyond the syntax borrowing more from C than Pascal, Ada 95 seems to have prefigured the entire Java-descended class of languages. https://en.wikipedia.org/wiki/Ada_%28programming_language%29#Control_structures ) |
| 20:30 | <&McMartin> | (That is, Ada looks more Pascal-y, Java looks more C-y) |
| 20:40 | <&McMartin> | Also, argh |
| 20:40 | <&McMartin> | This was-supposed-to-be-short post about offset assembly keeps wanting to be about DLLs and TLBs instead. |
| 20:42 | | Kindamoody is now known as Kindamoody|afk |
| 20:59 | | gizmore|2 [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has joined #code |
| 21:00 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has quit [Operation timed out] |
| 21:34 | | Kindamoody|afk is now known as Kindamoody |
| 21:45 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has joined #code |
| 21:47 | | gizmore|2 [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has quit [Ping timeout: 121 seconds] |
| 21:52 | | gizmore|2 [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has joined #code |
| 21:53 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has quit [Operation timed out] |
| 22:09 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has joined #code |
| 22:10 | | gizmore|2 [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has quit [Operation timed out] |
| 22:13 | | Kindamoody is now known as Kindamoody[zZz] |
| 22:14 | | gizmore [kvirc@Nightstar-af6i8r.dip0.t-ipconnect.de] has quit [Ping timeout: 121 seconds] |
| 22:40 | | ion_ [Owner@Nightstar-6grqph.vs.shawcable.net] has quit [Ping timeout: 121 seconds] |
| 22:40 | | ion [Owner@Nightstar-6grqph.vs.shawcable.net] has joined #code |
| 23:59 | | ErikMesoy [Erik@Nightstar-hq72t5.customer.cdi.no] has quit [[NS] Quit: Leaving.] |
| 23:59 | | ErikMesoy [Erik@Nightstar-hq72t5.customer.cdi.no] has joined #code |
| --- Log closed Sun Feb 14 00:00:32 2016 |