Mercurial > repo
view paste/paste.28875 @ 1810:50a542f0e95f
<oerjan> fetch http://oerjan.nvg.org/url
author | HackBot |
---|---|
date | Fri, 25 Jan 2013 20:27:04 +0000 |
parents | e037173e0012 |
children |
line wrap: on
line source
2008-05-09.txt:13:53:37: <ais523> I'm thinking of calling it Feather, because it's so lightweight compared to most Smalltalks 2008-05-09.txt:15:26:45: <ais523> hmm... I think Feather has to be lazy, to prevent it infinite-looping in loads of common cases 2008-05-12.txt:17:10:59: <ais523> In Feather, everything is a function 2008-05-12.txt:17:12:16: <ais523> whereas Feather truly has no pointers 2008-05-12.txt:17:13:38: <ais523> I don't know of a language that attempted to use time-travel to implement inheritance, though, before Feather 2008-05-12.txt:17:15:20: <ais523> in Feather, pass-by-value is pass-by-reference, there's no way to distinguish between them 2008-05-12.txt:17:16:10: <ais523> in Feather, there will be boxed types too 2008-05-12.txt:17:19:15: <ais523> at least from Feather's point of view 2008-05-12.txt:17:19:32: <ais523> the first thing a Feather program does when it starts up is to define its own primitives and syntax 2008-05-12.txt:17:20:12: <ais523> each iteration of the time loop happens simultaneously from Feather's point of view 2008-05-12.txt:17:21:35: <ais523> you just have to realise that Feather's timeline is not the same as the real-life timeline 2008-05-12.txt:17:22:05: <ais523> because Feather can be retroactively transformed into any other language 2008-05-12.txt:17:23:12: <ais523> but Feather itself is unaware that that's happened 2008-05-12.txt:17:24:08: <Deewiant> ais523: what part of this is different from "Feather is all programming languages, it just depends on the compiler/interp you use" 2008-05-12.txt:17:24:54: <Deewiant> ais523: so what rules does a Feather program have to obey to be able to be called that 2008-05-12.txt:17:25:27: <ais523> which is what makes it Feather rather than HQ9+, for instance 2008-05-12.txt:18:57:15: <ais523> it's only after thinking about Feather that I appreciate how it's even possible that a story can have someone go back in time, change a few details, then come back to the present and the world is much like it was when they left with only a few details changed 2008-05-12.txt:18:58:12: <ais523> both of those are common Feather scenarios, the second is something that you have to spend effort avoiding 2008-05-29.txt:20:26:25: <ais523> I've been thinking about writing a Feather bootstrap for a while now 2008-05-29.txt:20:26:54: <ais523> because I'm getting fed up of waiting for a feather interp to spontaneously come into existence 2008-05-29.txt:20:27:34: <AnMaster> ais523, feather? 2008-05-29.txt:20:35:59: <ais523> in Feather, assuming you're using a boxed number 2 (you would be normally but you need to deal with unboxed objects during bootstrapping) 2008-05-29.txt:20:36:32: <ais523> not everything is an object, unlike in Feather 2008-05-29.txt:20:55:58: <ehird> ais523: Feather, I hope. 2008-05-29.txt:20:56:47: <ais523> ehird: well, I have considered a feathernomic; its defining feature would probably be that all rulechanges were retroactive to the start of the nomic 2008-05-30.txt:20:55:52: <ais523> oh, Feather has no side effects either 2008-06-02.txt:18:44:59: <ais523> I have a question about what names I should use for things in Feather 2008-06-02.txt:18:53:30: <ais523> sorry, there aren't really in Feather either 2008-06-02.txt:18:54:54: <ais523> oh well, I'll have to think of something different in Feather 2008-06-02.txt:18:55:49: <ais523> (Feather's reminding me more and more of Smalltalk + Haskell + Scheme, by the way) 2008-06-02.txt:21:31:53: <ais523> Hiato: sounds like Feather in reverse 2008-06-02.txt:21:35:14: <ais523> but I've only got to documenting Basic Feather so far 2008-06-02.txt:21:50:04: <ais523> as for Feather, primitiveBe is the thing that really sets it apart from other languages, although you have to be careful using it to avoid an infinite loop 2008-06-03.txt:20:44:19: <ais523> anyway, I pasted a bit of what I'd done so far on Feather a while ago 2008-06-03.txt:20:45:45: <ais523> <ais523> anyway, I pasted a bit of what I'd done so far on Feather a while ago 2008-06-10.txt:17:33:46: <ais523> or like writing Feather in Feather? 2008-06-13.txt:19:20:08: <ais523> if there were, then maybe TRDS could be used to write Feather 2008-06-13.txt:19:21:38: <ais523> tusho: yes, that's the operation I want in Feather, really 2008-06-15.txt:17:18:22: <ais523> oh dear, this is starting to remind me of Feather again 2008-06-15.txt:18:01:33: <ais523> tusho: either that, or just wait for a Feather interp with the required features to spontaneously pop into existence 2008-06-15.txt:18:02:44: <ais523> whose only purpose was to optimise for the operations Feather did 2008-06-15.txt:18:03:17: <tusho> ais523: the good thing about feather in JS is that it'll be totally detached from traditional consoles 2008-06-15.txt:18:03:36: <ais523> the major unusual operation in Feather is giving an argument to a continuation which is almost identical to what the call/cc returned in the first place 2008-06-15.txt:18:06:20: <ais523> tusho: because say if you're writing a feather interp in JS 2008-06-15.txt:18:06:50: <ais523> anyway, a Feather interp in Feather is trivial 2008-06-15.txt:18:07:19: <ais523> Feather has eval so that you can modify it 2008-06-15.txt:18:08:40: <tusho> ais523: shall we just call feathejs ... feather? 2008-06-15.txt:18:09:03: <ais523> call it feathers 2008-06-15.txt:18:09:43: <tusho> ais523: feathers will just make people mix up feather and feathers 2008-06-15.txt:18:09:47: <ais523> I like the idea of multiple feather interps 2008-06-15.txt:18:10:11: <ais523> after all, feather images /are/ portable between interps 2008-06-15.txt:18:35:45: <ais523> but the class browser itself would be written in Feather 2008-06-15.txt:18:36:11: <ais523> tusho: you would make the jquery UI available as IO commands that Feather could use 2008-06-15.txt:18:37:16: <ais523> tusho: yes you can, you just reimplement it in Feather 2008-06-15.txt:18:37:30: <ais523> Feather is worse 2008-06-17.txt:22:05:06: <ais523> and things like Feather are just fun 2008-06-18.txt:15:36:31: <ais523> for instance, I'll definitely need a Feather interp before I figure out what the spec should be 2008-06-18.txt:15:42:01: <ais523> but the great thing about Feather is that it can be retrofitted onto the language 2008-06-18.txt:15:42:41: <ais523> Feather has eval so that you can modify it! 2008-06-18.txt:15:45:12: <ais523> oklofok: you can do that in Feather 2008-06-18.txt:15:45:37: <ais523> well, I'm trying to make the base of Feather as tarpitty as possible 2008-06-19.txt:16:09:01: <ais523> that retroactivity's so confused that it doesn't even make sense in Feather 2008-06-19.txt:16:13:28: <tusho> ais523: could you express it in feather, though? 2008-06-19.txt:16:13:48: <ais523> tusho: a retroactive Feather change has to be legal in the situation that was retroactively changed from 2008-06-19.txt:16:14:23: <ais523> tusho: nothing, it would be inexpressible, assuming that CANNOT in B Nomic == inexpressible in Feather 2008-06-19.txt:16:16:32: <ais523> oh, and a Feather compiler would have to bundle an interp 2008-06-19.txt:16:17:11: <ais523> e.g. you can get the source code for a Feather interp by first retroactively modifying the language to expose the source code of all functions, and then looking at it 2008-06-19.txt:16:31:12: <tusho> ais523: I think we need #feather, because js2cps is a pretty big thingy 2008-06-19.txt:16:31:28: <ais523> tusho: #feather already exists 2008-06-19.txt:16:32:51: <tusho> ais523: write an ircd in feather 2008-06-19.txt:16:33:09: <ais523> tusho: ugh, Feather hates IO 2008-06-19.txt:16:33:30: <tusho> ais523: it occurs to me that Feathejs is like the ideal implementation of feather 2008-06-19.txt:16:36:43: <ais523> #featherlang 2008-06-19.txt:16:36:48: <tusho> ais523: #feather-lang 2008-06-19.txt:16:43:16: <ais523> I'm also making a functional version, protoFeather 2008-06-19.txt:16:43:40: <ais523> and thus become Feather 2008-06-19.txt:16:43:51: <tusho> ais523: feathejs should implement protoFeather 2008-06-19.txt:16:50:38: <ais523> jix: finding ways to do that is one major part of my effort in Feather 2008-06-27.txt:18:07:05: <ais523> cherez: worth it, although Feather requires a whole new set of tenses 2008-06-27.txt:23:28:22: <ais523> psygnisfive: try implementing Feather, I'm struggling 2008-06-27.txt:23:29:47: <ais523> I have partial notes on a few feather-like objects 2008-06-27.txt:23:32:34: <ais523> AnMaster: Feather 2008-06-27.txt:23:38:46: <ais523> whereas in Feather you can take information from one possible path and use it to modify the next path that's tried 2008-06-27.txt:23:41:26: <ais523> also, feather's interesting due to consistent time travel rules 2008-06-27.txt:23:49:09: <ais523> oh, and the major problem with Feather programming is avoiding timeloops 2008-06-30.txt:23:32:44: <ais523> tusho: heh, you could do that in Feather 2008-07-03.txt:16:08:02: <ais523> olsner: TwoDucks (uncomputable), Feather (not properly specced and unimplemented) 2008-07-14.txt:20:45:47: <ais523> tusho: Feather has lambda too 2008-07-17.txt:21:15:49: <ais523> oklopol: I did, it reminded me a bit of Feather and a bit of TwoDucks 2008-07-17.txt:21:17:39: <ais523> oklopol: yes, I know, but that sort of lang is good for implementing Feather in 2008-08-06.txt:22:08:27: <ais523> in Feather, pass by reference and pass by value are indistinguishable 2008-08-06.txt:22:08:36: <SimonRC> ais523: where is Feather? 2008-08-06.txt:22:09:28: <ais523> Feather 2008-08-08.txt:20:44:18: <ais523> AnMaster: Feather is sufficiently general that after a while everything seems like "that sounds like Feather", it takes over your brain 2008-08-08.txt:20:44:48: <ais523> AnMaster: you absolutely need to bundle the source code with Feather, it's possible to prove it 2008-08-08.txt:20:45:18: <AnMaster> ais523, I don't know enough about feather really 2008-08-08.txt:20:45:47: <AnMaster> ais523, give us hello world in feather! 2008-08-08.txt:20:46:46: <ais523> AnMaster: no, although a Feathernomic would be awesome 2008-08-08.txt:20:47:21: <ais523> I really want to implement Feather 2008-08-08.txt:20:47:34: <tusho> ais523: let's implement feather starting tomorrow 2008-08-08.txt:20:47:34: <ais523> I'll start by implementing Protofeather, I think, which was a lang I invented to write a Feather interp in 2008-08-08.txt:20:50:00: <ais523> Feather's basic operation is reasonably easy to write in terms of call/cc 2008-08-08.txt:20:50:01: <AnMaster> ais523, well what is the basic syntax of non-modified feather? 2008-08-08.txt:20:50:43: <ais523> I invented Feather after thinking about Smalltalk for a while 2008-08-08.txt:21:59:48: <ais523> tusho: after Feather and Underload and Underlambda and all the other stuff we're going to do 2008-08-09.txt:16:25:42: <tusho> ais523: #feather-lang 2008-08-11.txt:19:49:45: <AnMaster> ais523, you never posted those specs and examples of feather btw 2008-08-23.txt:17:41:11: <AnMaster> ais523, feather? 2008-08-25.txt:17:17:58: <ais523> very Feather 2008-09-05.txt:20:06:30: <AnMaster> GreyKnight, like ais523's Feather? 2008-09-08.txt:20:05:33: <AnMaster> ais523, any news? feather maybe? 2008-09-12.txt:18:14:33: <AnMaster> ais523, how goes feather? 2008-09-16.txt:21:25:36: <AnMaster> ais523, how goes gcc-bf and Feather? 2008-09-20.txt:20:24:11: <ais523> "I am programming", in English, but "Feather is correct-to-program", in Latin literally translated to English 2008-09-23.txt:15:56:18: <AnMaster> ais523, Feather! 2008-09-23.txt:18:09:07: <AnMaster> ais523, and I would like Feather if you ever got it done ;P 2008-09-23.txt:18:12:35: <ais523> Feather would change everything 2008-09-23.txt:18:13:02: <AnMaster> ais523, could you even make any meaningful programs in Feather without changing the language? 2008-09-23.txt:18:15:34: <AnMaster> ais523, hello world program in feather please? 2008-09-23.txt:18:15:47: <ais523> AnMaster: I can't write anything in Feather yet, not even a NOP, really 2008-09-23.txt:18:16:20: <AnMaster> ais523, what language do you plan to implement feather in? or the boot strap feather in 2008-09-23.txt:18:18:28: <ais523> one of the main problems with Feather is I feel it needs an interp before it can be properly specced 2008-09-23.txt:18:30:00: <ais523> continuations are very very important to the way I plan to implement Feather 2008-09-23.txt:18:30:24: <ais523> in fact I was thinking about writing my own language in C, merely implemented to optimise continuations for the unusual use they'll see in a Feather program 2008-09-25.txt:22:18:16: <ais523> wow, oklofok invented Feather 2008-09-26.txt:21:20:59: <tusho> ais523: imagine an AnMaster written feather interp... 2008-09-29.txt:14:37:50: <ais523> unlike something like Feather which keeps running away when I think about it 2008-10-06.txt:09:49:47: <AnMaster> ais523, Feather? 2008-10-06.txt:09:54:56: <ais523> I plan to implement pretty much all of Feather with multireturning call/cc 2008-10-17.txt:12:39:47: <ais523> hmm... I've realised that all my 4 or so outstanding problems to solve before I can implement Feather are actually the same problem 2008-10-17.txt:12:40:56: <ais523> the outstanding problem in Feather is: to determine what in the program after a change corresponded to what before a change 2008-10-17.txt:14:56:22: <ais523> I think I might know how to get Feather to work... 2008-10-17.txt:15:00:05: <oklopol> ais523: sorry, i confused feather with your other unimplementable language 2008-10-17.txt:15:00:20: <ais523> Feather is at least in theory implementable, I think 2008-10-17.txt:15:14:45: <ais523> basically, the way I think Feather will do it is that input and output are tagged with their /purpose/ 2008-10-17.txt:15:14:56: <AnMaster> ais523, could you make feather become befunge? 2008-10-17.txt:15:15:21: <ais523> as at some point the program would need to be a Feather/Befunge polyglot 2008-10-17.txt:15:26:04: <AnMaster> ais523, could feather become any other language? 2008-10-17.txt:15:26:41: <ais523> unless you wrote the program as a Feather/Befunge polyglot from the start, which could be fun 2008-10-17.txt:15:26:52: <ais523> AnMaster: you'd have to implement them in Feather 2008-10-17.txt:15:27:11: <ais523> I/O and standard library and such couldn't be in part of the becomed language unless they were in Feather's stdlib to start with 2008-10-17.txt:15:28:42: <AnMaster> ais523, if you could have that on the first line as first char, the befunge/feather polygot would be easy 2008-10-17.txt:15:29:05: <ais523> and that also would make Befunge/Feather an easy polyglot 2008-10-17.txt:15:29:41: <ais523> AnMaster: it wouldn't care, at the time the Befunge program ran the Feather code would no longer exist 2008-10-17.txt:15:30:08: <ais523> if Feather modified itself into Befunge itself, you couldn't modify it back as Befunge has no command to retroactively change itself into Feather 2008-10-17.txt:15:30:42: <ais523> however, modifying Feather into Befunge definitely qualifies as an insane change 2008-10-17.txt:15:31:35: <ais523> the rule for a sane Feather parser change is that in addition to the other requirements to be sane, it has to be backwards-compatible 2008-10-17.txt:15:41:54: <ais523> hmm... you could do a Feather quine by causing the parser to output its input rather than parsing it, retroactively 2008-10-17.txt:15:45:10: <ais523> in Feather, though, the executable can modify the source and it changes accordingly, whilst still running 2008-10-17.txt:15:45:41: <ais523> hmm... lazy parsing might be useful in Feather, not necessary, but nice 2008-10-17.txt:15:47:06: <AnMaster> ais523, you could add it from within feather anyway ;P 2008-10-17.txt:15:47:47: <ais523> the nature of any Feather interp is that it has to be written entirely in Feather 2008-10-17.txt:15:48:18: <ais523> so you start off with two versions of the same interp, one in Feather, one in some other lang, which both do exactly the same thing 2008-10-17.txt:15:49:02: <ais523> that seems like a better bet than hoping a Feather interp will spontaneously come into existence, great as that would be 2008-10-17.txt:15:49:15: <ais523> once Feather is running, it's under the impression that there are an infinite number of layers of Feather interps under it 2008-10-17.txt:17:56:39: <ais523> AnMaster: not even if you write featherfunge some day? 2008-10-17.txt:17:57:24: <ais523> that involves having a working Feather first 2008-10-19.txt:12:02:24: <ais523> hmm... figuring out how to hot-change parts of a Feather program while it's running is a major problem 2008-10-20.txt:10:13:52: <ais523> oklopol: hmm... Feather makes me feel the same way, or worse 2008-10-20.txt:15:17:24: <AnMaster> then that is purely hypothetical, at least until ais523 implements Feather 2008-10-21.txt:18:56:01: <ais523|busy> I wonder if it will be easier or harder to implement than Feather 2008-10-21.txt:18:56:35: <ais523|busy> anyway, I think I've figured out how to prevent Feather going into an infinite loop (as opposed to arbitrary loop) when it parses the parser with itself 2008-10-21.txt:18:57:04: <ais523|busy> Deewiant: it's a Feather thing 2008-10-21.txt:19:03:08: <ais523|busy> #feather-lang is the appropriate channel 2008-11-03.txt:17:42:02: <ais523> ehird: please don't randomly throw out Feather references to people who haven't seen it before, it took several weeks to explain the first time 2008-11-03.txt:17:47:11: <ais523> for what, Feather? 2008-11-03.txt:17:47:29: <ais523> so the parser, and indeed the entire interp, needs to be written from scratch in Feather 2008-11-03.txt:17:51:09: <ais523> which is about the only meaningful operation in Feather, and also the one that causes all the headaches 2008-11-03.txt:17:51:46: <ais523> a Feather codenomic would be entirely platonic, and you could retroactively change what the rules were in the past and have everything reinterpreted under those rules 2008-11-03.txt:17:52:46: <ais523> the main problem is that this is how Feather handles /everything/ 2008-11-03.txt:17:53:26: <ais523> Feather defies comparison to anything, really 2008-11-03.txt:18:04:30: <ais523> I invented Feather originally because Smalltalk wasn't Smalltalky enough IMO, but it ended up as something moderately different 2008-11-03.txt:18:37:45: <ais523> unlike Feather, Cyclexa does have a spec 2008-11-03.txt:18:37:52: <ais523> but like Feather, it doesn't have an implementation 2008-11-05.txt:19:47:18: <ais523> the main problematicness in continuation-related thinking processes is the go-back-in-time thing (btw, I'm planning to use them for Feather, they're the perfect choice for it) 2008-11-10.txt:20:05:59: <AnMaster> ais523, how goes gcc-bf? ick? feather? 2008-11-14.txt:22:16:34: <AnMaster> ais523, how goes ick, gcc-bf, feather and so on? 2008-11-25.txt:15:51:12: <ais523> ehird: Feather 2008-12-22.txt:18:50:44: <ais523> ehird: stop inventing Feather 2009-01-14.txt:19:56:47: <AnMaster> <ais523> because then you could change the syntax at runtime <-- Feather! 2009-01-20.txt:18:02:51: <AnMaster> ais523, what about feather? It could be both :P 2009-01-20.txt:18:02:58: <ais523> AnMaster: ah, Feather 2009-01-20.txt:18:08:35: <ais523> incidentally, my guiding goal for Feather starting syntax is "looks vaguely like Smalltalk, but for different reasons" 2009-01-20.txt:18:09:00: <AnMaster> ais523, oh, interesting, I would never have ended up with feather by that 2009-01-22.txt:15:21:10: <ais523> my other big new esolang project, besides Feather 2009-02-13.txt:21:49:18: * ais523 feather-dusters oerjan ----<<< 2009-02-15.txt:17:13:13: <ais523> oerjan: there's the method Feather uses 2009-03-01.txt:18:34:53: <AnMaster> ais523, any progress on Feather? 2009-03-08.txt:18:52:30: <ais523> or Feather and get people shouting at me to make some progress 2009-04-01.txt:23:04:26: <AnMaster> ais523, could you write Proud be written in Feather? 2009-04-01.txt:23:04:38: <ais523> no, Proud's uncomputable, Feather is not super-TC 2009-04-01.txt:23:04:46: <ais523> although Feather hurts my head, and I don't want my brain to explode right now 2009-04-01.txt:23:06:05: <ais523> cpt_obvious: I suggest you ask in #feather-lang, I think it's empty atm so it'll be safe to ask 2009-04-01.txt:23:07:00: <ais523> which is another reason not to explain Feather right now 2009-04-01.txt:23:07:12: <ais523> explaining Feather to someone /who is drunk/ probably would cause a fatality 2009-04-01.txt:23:07:12: <ehird> ais523: I don't know; alcohol may well improve perception of Feather 2009-04-01.txt:23:09:05: <ais523> seeing as nothing in Feather can ever change 2009-04-05.txt:12:25:40: <AnMaster_ipv6> ais523, I guess you could do it in Feather though? 2009-04-05.txt:12:25:57: <ais523> Feather doesn't really define its I/O environment at all 2009-04-09.txt:22:31:58: <AnMaster> ais523, Feather is TC to parse right 2009-04-09.txt:22:39:26: <ais523> AnMaster: what I mean with Feather is that the first pass is always trivial, because it's just tokenising into letters 2009-04-09.txt:22:44:18: <AnMaster> ais523, write a first stage parser then. I mean do you ever plan to spec, or even implement feather? 2009-04-26.txt:19:08:14: <AnMaster> ais523, any progress on Feather 2009-05-28.txt:01:57:00: <coppro> ais523: what do the silver/gold medals and feathers on the level selection screen mean? 2009-05-28.txt:01:57:30: <ais523> coppro: feather = unsolved easy level, silver = solved easy level, gold = hard level 2009-05-28.txt:01:57:38: <ais523> you can click on the feather button to switch between easy and hard mode 2009-05-29.txt:01:06:04: <ais523> ehird: well, #feather-lang exists 2009-07-07.txt:22:03:10: <AnMaster> ais523, I invite you to #feather-lang btw :P 2009-07-10.txt:15:35:52: <AnMaster> ais523, any progress on feather? 2009-07-10.txt:15:35:59: <ais523> AnMaster: ducking a feather is easy 2009-07-10.txt:15:36:05: <ais523> or maybe it's a duck feather? 2009-08-14.txt:23:02:38: <ais523> heh, he's still in #feather-lang 2009-08-14.txt:23:04:55: <AnMaster> ais523, wow you missed lots of action in #feather-lang 2009-08-14.txt:23:05:57: <ais523> was it Feather-related? 2009-08-26.txt:18:09:16: * ais523 considers using a Feather-OS, and hotswapping init 2009-09-17.txt:20:43:24: <AnMaster> ais523, like Feather? XD 2009-09-29.txt:10:06:25: <ais523> start with a minimal interpreter that's enough to 'be Feather' in that it can be retroactively self-modified 2009-09-29.txt:10:09:03: <ais523> because it wouldn't need to be written in Feather 2009-09-29.txt:10:09:05: <ais523> that isn't an "of course" with Feather 2009-09-29.txt:10:10:14: <AnMaster> ais523, it was "of course the outermost _interpreter_ can't be in Feather itself" 2009-09-29.txt:10:10:14: <ais523> for a while I thought it had to be a Feather/something else polyglot 2009-09-29.txt:10:10:32: <ais523> but it turns out that you can see the source code of an interp even if it isn't written in Feather 2009-09-29.txt:10:24:04: <ais523_> and second, it should be easy to write a polyglot in that lang and Feather 2009-09-29.txt:16:20:38: <ais523> Feather is /actually/ driving me mad 2009-09-30.txt:17:03:28: <ais523> and if it's opaque, all Feather programs will need to be quines 2009-09-30.txt:18:46:50: <ais523> I mean, I was talking about Feather earlier 2009-09-30.txt:18:47:04: <ais523> partly because thinking about Feather drives everyone mad, I suppose 2009-09-30.txt:18:51:26: <AnMaster> ais523, is it possible to express paradoxes in feather? 2009-09-30.txt:18:56:14: <deschutron> ais523: your feather idea sounds interesting, though i don't know whether it provides any convenience at all 2009-09-30.txt:19:00:41: <ais523> deschutron: same thing, no paradox under Feather's rules 2009-09-30.txt:19:03:06: <ais523> everything is immutable in Feather 2009-09-30.txt:19:03:37: <ais523> ehird: I see what you're saying; I think that's the Feather equivalent of a trampoline, or something like taht 2009-09-30.txt:19:20:40: <ais523> worryingly, I think you probably /have/ to program in that sort of way, and that's what makes Feather an esolang 2009-09-30.txt:19:22:25: <ais523> it's not unlike Feather, in a way 2009-09-30.txt:19:23:09: <ais523> deschutron: I/O is a real pain in Feather, for this sort of reason 2009-09-30.txt:19:24:40: <ais523> deschutron: well, the language is called Feather because the language itself is bare-bones 2009-10-01.txt:14:28:24: <ais523> conclusion: I'm thinking too much about Feather 2009-10-03.txt:19:58:01: <ais523> deschutron: oh, I saw your quasi-Feather thing, it's rather unlike Feather, but ofc that doesn't mean it's a bad lang 2009-10-03.txt:19:58:12: <ais523> I think pretty much every entity in existence is rather unlike Feather 2009-10-03.txt:20:13:58: <deschutron> its a language I thought of after talking to ais523 about feather 2009-10-05.txt:17:49:00: <ais523> actually... I think the way it works is, that the interp isn't originally written in Feather 2009-10-05.txt:17:49:08: <ais523> such that the interp ends up mostly in Feather 2009-10-05.txt:18:03:08: <ais523> "Feather is a language with meany time travel paradoxes!" 2009-10-05.txt:18:05:00: <ais523> "(Cheat slightly; you know Feather so pick out the relevant facts)" 2009-10-05.txt:20:06:04: * ais523 wonders whether to monologue about Feather, ignoring what the rest of the channel says 2009-10-05.txt:20:09:03: <AnMaster> ais523, and I'm interested in garli^Wfeather 2009-10-07.txt:15:44:30: <AnMaster> ais523, started coding on the feather implementation? 2009-10-07.txt:15:50:42: <ais523> hmm... Feather's the only language I know in which you have to worry about portability within a program 2009-10-09.txt:15:05:54: <ais523> incidentally, I had a go at writing a Feather interpreter 2009-10-09.txt:15:06:54: <ais523> the issue is that causality loops are all very well in Feather, but tend not to work in Scheme 2009-10-09.txt:15:08:15: <ehird> ais523: wild speculation - the only way to run Feather programs acceptably will be via a custom-written C vm, because using retroactivity blows up supermegaexponentially when using scheme 2009-10-09.txt:15:09:11: <ais523> I have crazy ideas of optimising Feather interps which involve using diff-based call/cc 2009-10-09.txt:15:10:27: <ais523> in fact, I now have something like three stages of Feather interpretation 2009-10-09.txt:15:10:51: <ais523> a Feather seed (what I'm writing, it is something that doesn't obey all the rules of Feather, but /can/ be retroactively modified so it does) 2009-10-09.txt:15:11:02: <ais523> a Feather kernel (something that implements a subset of Feather sufficient to bootstrap to the whole thing) 2009-10-09.txt:15:11:11: <ais523> and a Feather interpreter (which implements the whole thing) 2009-10-19.txt:18:26:23: <AnMaster> ais523, feather progress? 2009-10-19.txt:18:27:45: <AnMaster> ais523, what from? feather? 2009-10-19.txt:18:33:21: <ais523> Feather has really messed-up causality, which is really frustrating my attempts to write an interp for it 2009-10-24.txt:22:34:45: <AnMaster> ais523, start thinking about feather please 2009-10-28.txt:22:31:56: <ais523> incidentally, this conversation is a good reflection on Feather 2009-10-28.txt:22:32:29: <AnMaster> ais523, how is it a good reflection on Feather? 2009-10-28.txt:22:32:48: <ais523> Feather manages to simulate that 2009-10-31.txt:19:54:20: <AnMaster> ais523, how goes Feather? 2009-11-16.txt:21:50:14: <ais523> anyway, I at least came up with snappy names for Feather's assignment and clone operators 2009-12-01.txt:20:55:55: <ais523> not finished, and unlikely to be for a while due to RL pressures, and not being top of my esolang priorities (Feather and Underlambda are higher) 2009-12-08.txt:10:23:46: <AnMaster> ais523, oh and I invented a feather-like language 2009-12-08.txt:10:24:37: <ais523> AnMaster: go on, although I doubt feather-like is very easy to achieve at all without being utterly different 2009-12-29.txt:15:25:17: <AnMaster> ais523, how goes stuff with feather? 2009-12-29.txt:15:26:14: <ais523> I'd /almost/ even forgotten Feather existed... 2009-12-30.txt:03:41:11: <ais523> pikhq: you can only imagine how hard it is in Feather 2009-12-30.txt:06:39:34: <ehirdiphone> ais523: maybe dependent types are the key to feather :P 2010-01-10.txt:17:17:30: <ais523> it's correct if used as "ais523, could you please go and implement Feather for me RIGHT NOW?" 2010-03-09.txt:12:30:40: <alise> ais523: about Feather. 2010-03-09.txt:12:31:12: <alise> ais523: What is Feather, in detail? 2010-03-09.txt:12:36:43: <ais523> alise: planned Feather syntax is [ x | f x ] 2010-03-09.txt:12:42:42: <ais523> anyway, anything remotely related to Feather is inherently weird 2010-03-09.txt:15:00:03: <AnMaster> <ais523> alise: planned Feather syntax is [ x | f x ] <-- yay Feather! 2010-03-09.txt:15:53:39: <alise__> ais523: Hey, it's for Feather. 2010-03-09.txt:16:22:28: <alise> ais523: so, do you know what the bootstrap progam will look like for feather? 2010-03-09.txt:16:24:34: <ais523> alise: the issue is making them forward-compatible with actual Feather atoms 2010-03-09.txt:16:32:29: <ais523> that's the start of an attempt to implement Feather 2010-03-09.txt:17:39:57: <cpressey> Thank you ais523, I am now exactly as confused about feather as I was before. 2010-03-09.txt:17:43:31: <ais523> cpressey: that's an interesting analogy; and about as close to the truth as anything else about Feather that I've managed to put into words 2010-03-22.txt:13:51:00: <ais523> (except in a few cases like Feather, which not even I understand and I thought up the idea...) 2010-03-31.txt:00:20:18: <ais523> (it wouldn't be in Perl6 or C++, it would be in Lua, JS and Feather) 2010-03-31.txt:00:21:17: <ais523> and there'll probably be some way to do that sort of thing in Feather, although single inheritance is easier there than multipl 2010-03-31.txt:00:28:26: <ais523> hmm, you can do that in Feather too 2010-04-16.txt:21:25:14: <ais523> Sgeo_: try imagining a cross between Prolog and Feather, then fail 2010-04-23.txt:19:43:19: <AnMaster> ais523, any progress at with feather btw? (It was weeks since I asked last time!) 2010-06-07.txt:17:59:20: <AnMaster> ais523, after you finish feather and gcc-bf that is ;P 2010-06-15.txt:17:00:01: <AnMaster> ais523, so do you think you will have any time for feather, gcc-bf and so on during the summer? 2010-06-20.txt:23:13:50: <ais523> must... not... mention... Feather.... 2010-06-20.txt:23:18:45: <AnMaster> <ais523> must... not... mention... Feather.... <-- ooh feather! 2010-06-23.txt:16:02:44: <Phantom_Hoover> ais523, what's this feather thing I've heard about? 2010-06-23.txt:16:18:06: <ais523> anyway, Feather's going to get a worse reputation than MAGENTA at this rate 2010-06-23.txt:19:22:39: <ais523> cpressey: that happens with Feather too 2010-06-25.txt:13:57:20: <Phantom_Hoover> ais523, maybe you can if you invent Feather? 2010-06-25.txt:13:57:45: <ais523> Phantom_Hoover: Feather programs can't retroactively change events outside the program 2010-06-25.txt:14:06:57: <ais523> although if I ever get around to writing a Feather standard library, a memory of input is definitely one of the things that would be useful [too many lines; stopping]