00:14:10 <lament> this isn't a city deep under the sea?
00:35:49 <ehird`> so! who wants to help work on the underload->C compiler
00:58:18 -!- skabet has joined.
01:05:57 -!- puzzlet_ has quit (Remote closed the connection).
01:05:57 -!- puzzlet has joined.
01:27:13 <oklopol> ehird`: written in python?
01:43:33 -!- ehird` has quit.
01:48:44 -!- puzzlet has quit (Remote closed the connection).
01:48:45 -!- puzzlet_ has joined.
02:16:15 -!- skabet has left (?).
03:46:22 -!- puzzlet has joined.
03:46:25 -!- puzzlet_ has quit (Remote closed the connection).
04:40:10 -!- GreaseMonkey has quit ("Hi Im a qit msg virus. Pls rplce ur old qit msg wit tis 1 & hlp me tk ovr th wrld of IRC. and dlte ur files. thx.").
05:21:21 -!- puzzlet_ has joined.
05:31:41 -!- puzzlet has quit (Read error: 110 (Connection timed out)).
06:26:18 -!- Justin_ has joined.
06:26:39 -!- Justin_ has left (?).
07:59:59 -!- clog has quit (ended).
08:00:00 -!- clog has joined.
08:08:52 <oklopol> http://www.vjn.fi/oklopol/graphica.txt turns out it *was* trivial to make an n-dimensional hypercube
08:09:17 <oklopol> also turns out my implementation is slow as hell :D
08:42:31 -!- oklopol has changed nick to Z.
08:42:46 -!- Z has changed nick to oklopol.
08:50:26 -!- oerjan has joined.
09:44:09 <Slereah> Am I supposed to mod this by 94 to get hello?
09:45:48 <oklopol> i should ask you that very same question.
09:47:59 -!- oerjan has quit ("leaving").
09:48:42 -!- slereah_ has joined.
10:04:27 -!- Slereah has quit (Read error: 110 (Connection timed out)).
10:48:34 -!- puzzlet_ has quit (Remote closed the connection).
10:48:46 -!- puzzlet has joined.
11:27:12 -!- puzzlet has quit (Remote closed the connection).
11:27:19 -!- puzzlet has joined.
11:47:29 -!- puzzlet has quit (Remote closed the connection).
11:47:35 -!- puzzlet has joined.
13:00:24 * slereah_ will try to implement the 2D version of the Love Machine 9000.
13:07:29 -!- jix has joined.
13:24:11 -!- Corun has joined.
13:42:49 -!- helios24 has joined.
13:47:12 -!- sebbu has quit ("@+").
14:01:30 -!- sebbu has joined.
14:03:00 -!- puzzlet has quit (Remote closed the connection).
14:03:02 -!- puzzlet has joined.
14:14:54 -!- Corun has quit ("This computer has gone to sleep").
14:36:02 -!- Corun has joined.
14:38:17 -!- slereah__ has joined.
14:38:17 -!- slereah_ has quit (Read error: 104 (Connection reset by peer)).
14:40:23 -!- Corun has quit (Client Quit).
14:56:45 -!- slereah__ has quit (Read error: 104 (Connection reset by peer)).
14:56:48 -!- slereah_ has joined.
15:08:21 -!- helios24 has quit (Remote closed the connection).
15:34:46 -!- ehird` has joined.
15:40:09 -!- jix has quit (Read error: 104 (Connection reset by peer)).
15:42:39 -!- timotiis has joined.
15:44:50 <ehird`> who is fast with log grepping?
15:45:00 <ehird`> i need to know the date where me and ais523 discussed 'turing complete regexps'
15:49:00 <Asztal> http://tunes.org/~nef/logs/esoteric/07.10.03
15:49:06 <Asztal> or maybe http://bespin.org/~nef/logs/esoteric/07.11.12
15:49:28 -!- jix has joined.
15:49:30 <Asztal> or maybe neither, btw, I used http://www.google.co.uk/search?hl=en&q=ais523+ehird+turing+complete+regexp :D
15:56:14 <ehird`> 2007-10-03 is right, Asztal
15:57:03 <slereah_> The 2D Love Machine has some problems dealing with the adding-new-squares-around
15:58:16 -!- helios24 has joined.
16:00:01 * ehird` (a(b)) -> [['a', ['b']]]
16:02:04 <ehird`> <ehird`>FUCK! I'm starting to think like Larry Wall.
16:02:09 <ehird`> wait, why am i still ehird`
16:02:10 -!- ehird` has changed nick to ehird.
16:12:48 -!- fax has joined.
16:15:33 <fax> I found a great Logic textbook
16:16:18 <fax> Lectures on the Curry Howard Isomorphism
16:30:25 <slereah_> http://membres.lycos.fr/bewulf/Russell/Hello.png
16:30:57 <slereah_> There seem to be some problems with some of the tape-displaying functions.
16:31:12 <slereah_> But then again, it's not the most important part I guess.
16:31:49 <slereah_> One day, I'll have to dig in that code to clean it. I don't look forward to it.
16:33:44 <ehird> your code looks like random line noise.
16:34:35 <slereah_> Well, it's just directions and "print space".
16:35:16 -!- sebbu2 has joined.
16:39:18 <slereah_> http://membres.lycos.fr/bewulf/Russell/Hello2.png
16:39:31 <slereah_> Hides the problems better with space as the empty char!
16:41:37 -!- ais523 has joined.
16:42:20 -!- ais523 has set topic: What is the title of this topic? Logs: http://tunes.org/~nef/logs/esoteric.
16:42:35 <ais523> Freenode get annoyed if a channel is publically logged but you don't tell anyone
16:43:03 <ais523> wow, ehird, you got your nick back?
16:43:35 <ehird> I also have 'rice'
16:43:39 <ais523> and as for those TC regexps, I'm still working on them
16:43:49 <ais523> written about 70% of the spec, now
16:43:50 <ehird> because, i saw an unused-for-a-year, short, memorable dictionary word, and nabbed it
16:43:56 <ais523> and started working on the implementation for a third time
16:43:57 <ehird> i'm picking mine up again
16:44:01 <ehird> kind of from the start
16:44:17 <ais523> now I have a version of Perl that understands (??{$regexp})
16:44:28 <ehird> ais523: Why embed it in perl?
16:44:31 <ehird> It's turing complete by itself!
16:44:40 <ais523> I'm using Perl to write the interpreter
16:44:57 <ais523> but without that construct, matching brackets is tedious
16:45:13 <ais523> of course, it exists in my regexp language too
16:45:18 <ais523> with a much shorter notation
16:45:50 <ehird> Mine will be written in Haskell or something, probably.
16:46:13 <ais523> one of my favourite innovations is a concise way to do yacc-like matching
16:46:19 <ehird> ais523: what's a^nb^nc^n
16:47:21 <ais523> or you can use :- rather than :1 if you don't want to hardcode a group number
16:47:21 <ehird> that has changed since last time..
16:47:34 <ais523> what did I say last time?
16:48:29 <ais523> <ais523> also, I have things like (a*)b*:-c*:- to solve the famous n as, n bs, n cs problem, which I think is considerably shorter than the corresponding BF program
16:48:39 <ais523> so it's the same program
16:48:49 <ais523> except that I didn't put the ^ and $ to anchor it to the whole expression
16:48:57 <ais523> incidentally, ^ and $ have completely different implementations
16:50:00 <ehird> i'm going to start with regexps as i like them, then extend to matching brackets
16:50:08 <ehird> ais523: does that sound like a good basis?
16:50:16 <ais523> mine is also based on recursion
16:50:27 -!- sebbu has quit (Connection timed out).
16:50:30 <ais523> a* is just an abbreviation for (|a$+)
16:50:50 <ais523> i.e. either match nothing, or match a followed by a repeat of the current group's code
16:51:08 <ais523> strictly speaking, (??|a$+), but the ?? just prevents the group being allocated a number
16:51:24 <ehird> I want a lambda symbol
16:51:39 <ais523> lambda in a RE language
16:51:58 <ehird> \(\)|\($<lambda>\)
16:52:00 <ais523> it would be easy enough in a self-modifyin language
16:52:04 <ehird> matches balanced brackets, but does nothing with them
16:52:14 <ehird> \(\)()|\(($<lambda>)\)
16:52:19 <ehird> matches brackets into a parse tree
16:52:25 -!- puzzlet_ has joined.
16:52:25 -!- puzzlet has quit (Remote closed the connection).
16:52:40 <ais523> my language produces a parse tree no matter what you do
16:52:49 <ais523> and the colons force two regexps to have the same parse tree
16:53:39 <ais523> you can choose whether . matches just a character, or whether it's equivalent to (\x0|\x1|\x2|...|a|b|c|...|\x10fffe|\x10ffff) using options
16:53:49 <ais523> (or whatever Unicode goes up to nowadays)
16:54:05 <ais523> and which option in a set is chosen is also taken into account
16:54:13 <ehird> ais523: does that account for all the nomansland stuff in unicode?
16:54:18 <ehird> those higher 'planes' or whatever they are
16:54:26 <ehird> it's all very epic and spooky
16:54:34 <ais523> all I know is that there are some codepoints that can't be used
16:54:37 <ehird> 'The Unicode encoding space consists of 17 planes with each plane consisting of 65536 code points'
16:54:52 <ais523> that at least explains where the 10ffff comes from
16:54:54 <ehird> As of Unicode 5.0.0, 102,012 (9.2%) of these code points are assigned, with another 137,468 (12.3%) reserved for private use, 2,048 for surrogates, and 66 designated noncharacters, leaving 872,582 (78.3%) unassigned. The number of assigned code points is made up as follows:
16:55:30 <ais523> I wonder when if ever they'll run out and have to add more planes?
16:55:35 <ais523> the BMP itself is sufficient for most things
16:56:25 <ehird> http://en.wikibooks.org/wiki/Windows_Programming/Unicode/Character_reference/E0000-E0FFF do they really need this page.
16:56:33 <ais523> you can also do things like ('a-z=A-Z')
16:56:39 <ais523> to change lowercase to caps
16:56:59 <ais523> and as for do they really need that page; I think that might have beeen on Wikipedia once but they moved it because Wikibooks was a better fit
16:57:13 <ais523> and it seems reasonable that some book might need page after page of Unicode characters in it...
16:57:38 * ais523 has to go for a while, but should be back maybe about half an hour later
16:57:56 -!- ais523 has quit ("hopefully back in about 1e mins...").
17:05:57 -!- slereah_ has quit (Read error: 104 (Connection reset by peer)).
17:06:14 -!- slereah_ has joined.
17:06:50 -!- slereah_ has changed nick to Slereah.
17:35:24 -!- ais523 has joined.
17:45:49 -!- fax has changed nick to afx.
17:50:01 -!- RedDak has joined.
17:54:36 -!- afx has changed nick to fax.
17:56:33 <ais523> ^(\`$+$+=$+$+\^|'skivdce(\..)')$
17:57:58 <ais523> that converts Unlambda from Polish to Reverse Polish notation
17:58:24 <ais523> it's easy enough to expand on it to produce an Unlambda to Underload converter in a single statement, except for c
18:00:48 <ais523> of course, my implementation isn't finished yet, so I haven't tested any programs
18:00:59 <ais523> I've only got as far as implementing the crazy quoting and precedence rules
18:01:24 <ehird> i'm still thinkning about mine
18:01:37 <ehird> ais523: i think i'll start a rewrite of the compiler's scheme code
18:01:41 <ehird> the prelude staying the same ofc
18:01:51 <ehird> i was thinking, I dunno. Haskell? It'd be good for writing the optimizations..
18:01:57 <ais523> I'm better at the C than at the Scheme
18:02:08 <ehird> yeah, and the scheme is ugly
18:02:11 <ais523> but I think the C's about as good as it's likely to easily get
18:02:14 <ehird> which is why i'm switching to something else :)
18:02:25 <ehird> (one of the things i want to add is a real parse tree)
18:02:30 <ais523> I figured out how to make dup O(1), but the other costs of doing so are sufficiently high that it isn't worth it
18:02:35 <ehird> also, with a new frontend
18:02:38 <ehird> we can optimize the code
18:02:50 <ehird> ais523: so, what languages do you understand? haskell might be a bit optimistic i guess :)
18:03:02 <ais523> I'm trying to learn it without docs
18:03:05 <ais523> which is a bit difficult
18:03:16 <ais523> and I'm not on my own computer at the moment, so I don't have access to an interpreter anyway
18:03:22 <ais523> nor to svn for that matter
18:04:36 <ais523> my computer is currently over 4 miles away from my Internet connection...
18:05:54 <ehird> ais523: this seems like a bit of a problem ;)
18:06:08 <ais523> well, I'm at the Internet connection at the moment, obviously
18:06:28 <ehird> but not your computer.
18:06:49 <ais523> no, so I'm stuck with the ancient version of ChatZilla
18:09:04 <ehird> ais523: do you even have a decent editor? :-)
18:09:31 <ais523> when I connect to Unix, like I do now, I have Emacs or vi
18:09:41 <ais523> otherwise, Visual Studio, Notepad, Wordpad and Textpad
18:09:49 <ais523> (and MS Office, of course, bu I don't count that)
18:09:57 <ais523> it's your opinion as to what's a decent editor
18:10:07 <ais523> personally I find sed works quite well in many cases
18:11:24 <ehird> wellll, let's see..
18:11:41 <ehird> emacs, blah, blah, blah, blah ... ok, you have one good OS which has a good editor inside
18:11:50 <ehird> if you're connecting to unix already that's an awful lot of indirection :-)
18:12:07 <ais523> it's worse; some of the projects I do are on a friend's Linux server
18:12:16 <ais523> and I have to connect to Unix to ssh over to Linux...
18:12:26 <ais523> it's great fun trying to run curses programs like that
18:12:35 * ais523 uses Linux at home, by the way
18:12:59 <ehird> haha, wow.. I telnetted in to Franz's Allego Lisp demo prompt for no reason
18:13:06 <ehird> just like the good ol' days
18:13:26 <ais523> <backspace> = <left arrow on keyboard>
18:13:37 <ais523> but the backspace key worked as well, fortunately for everyone's sanity
18:13:56 <ehird> ais523: still in windows prompts today..
18:13:59 <ehird> or at least: xp era.
18:14:15 <ehird> I should netcat the whole swank source to the prompt
18:14:17 <ehird> and conncet with SLIME
18:14:17 <ais523> run doskey and it gets a lot saner
18:14:29 <ais523> what, you mean netcat | sh
18:14:42 <ehird> you know what SLIME/swank is
18:14:56 <ehird> SLIME = Superior Lisp Interaction Mode For Emacs
18:15:11 <ehird> basically, it's what Symbolics would have been proud of.
18:15:13 <ais523> is that real lisp, or elisp?
18:15:23 <ehird> The whole editor gets knowledge about your lisp system
18:15:35 <ehird> SLIME actually connects with a socket to your lisp implementation
18:15:36 <ais523> Emacs knows scarily much anyway
18:15:39 <ehird> which is running swank
18:15:42 <ehird> swank is the backend
18:15:47 <ehird> with some implementation-specific parts
18:15:53 <ehird> and it tells SLIME /everythiing/ it wants to know
18:16:20 <ehird> from basic things like 'complete based on symbol table' to crazy, crazy introspection
18:16:25 <ehird> and it has a lisp mode
18:16:31 <ehird> but vastly superior
18:16:40 <ehird> with debugger integration and everything
18:16:50 <ehird> it's one of the Must Haves for CL development
18:16:59 <ais523> Emacs has all those features which are great but nobody who was brought up on a GUI will like or even understand
18:17:03 <ehird> but anyway... i should cat that over to franz' telnet and get SLIME on it
18:17:09 <ehird> ais523: I was brought up on a gui..
18:17:22 <ais523> but you managed to shake yourself out of it
18:17:30 <ehird> i'm still running a gui irc client here
18:17:38 <ais523> connecting to the Internet doesn't count
18:17:42 <ehird> and - shock! - using a mouse!
18:17:46 <ais523> it became designed for a GUI
18:17:55 <ehird> ais523: occasionally i end up using the mouse in emacs :P
18:18:00 <ais523> and actually I don't use mice all that often nowadays, now I have a laptop
18:18:00 <ehird> i'm fast with a mouse though
18:18:02 <ehird> so it doesn't matter
18:18:26 <ehird> my editing style is really idiosyncratic
18:18:37 <ais523> so is most people's, I expect
18:18:40 <ehird> it comes from using every editor from Notepad to vim to TextMate to ...
18:19:29 <ehird> it overwhelms you with pure awesomeness, too bad it can't really do shit i guess
18:19:36 <ehird> actually i have al lthe plan9 tools installed on here
18:19:43 <ehird> including an x11 acme
18:19:49 <ehird> ais523: OH! Have you used sam?
18:19:52 <ehird> another plan9 invention
18:19:59 <ehird> a top frame, with an ed prompt
18:20:06 <ehird> and loads of windows below each with one file
18:20:18 <ehird> and you can edit the file with all the flexibility of ed, while seeing it below
18:20:24 <ehird> it's even GRAPHICAL!
18:20:31 <ais523> sort of like the relation of ddd to gdb?
18:20:43 <ais523> I was started on ddd, but grew to prefer gdb
18:20:50 <ais523> ddd is sort of a GUI for gdb
18:21:14 <ais523> it runs gdb in a separate process, and translates GUI input into gdb commands and gdb output into graphical displays
18:21:26 <ais523> and you can see the gdb commands scrolling past at the bottom
18:22:27 <ehird> ais523: plan9 is crazy
18:22:27 <ais523> anyway, this /is/ the computer on which I wrote my buggy elisp abstraction eliminator
18:22:30 <ehird> ever used it's WM?
18:22:40 <ais523> so I'm pasting it now so I can later get it onto my laptop
18:23:09 <ais523> and no, I've only used Win 3.1 and Win 95 onwards WM, X11, the Common Desktop Environment, Gnome and KDE
18:23:30 <ais523> not on my own computer
18:23:36 <ais523> I've used it on other computers
18:23:43 <ais523> and I'm a fan of several of its features
18:23:52 <ais523> it's probably slightly better in UI terms than Gnome/KDE, but not by much
18:24:05 <ehird> it is once you learn about it through using it
18:24:30 <ehird> i must admit that i'm an expos addict, though mostly keyboard-based :P
18:24:55 <ais523> anyway, the abstraction eliminator is at http://pastebin.ca/860539
18:25:03 <ais523> it's buggy if you don't use a prefix arg to disable optimisation, though
18:25:18 <ais523> I wrote it so long ago that I've forgotten the syntax, though so I'd have to work it out again
18:25:50 <Slereah> Is it for lambda expressions in Haskell syntax?
18:26:07 <ais523> with an extension to write lambdas
18:26:32 <ais523> it seems to be ^ and $, according to the source
18:26:42 <ais523> ^ to declare a lambda binding, $ to use it
18:28:06 <ehird> FONT LOCK WHAT THE HELL
18:28:07 <ais523> I make it a habit to write Emacs modes for all the esolangs I work in
18:28:31 <ais523> only intercal-mode is publically released, though, and I didn't even originally write that
18:28:39 <ais523> just added font lock and some helpful functions to it
18:29:26 <ais523> which come to think of it is actually wrong
18:29:47 * Slereah uses this one : http://membres.lycos.fr/bewulf/Russell/Lazy%20Bird/ae9.py
18:29:48 <ais523> because the input should have been `^x`$x$x^x`$x$x
18:30:01 <ais523> giving the correct ```sii``sii
18:30:11 <Slereah> Although it doesn't differentiate combinators from variables.
18:30:25 <Slereah> So s, k, i, c and b are better left out of the expression.
18:30:26 <ehird> I still th ink we need an abstraction introducer.
18:30:43 <ehird> ```sii``sii -> ((\x. x x) (\x. x x))
18:30:49 <ais523> who was it who said that decompiling Unlambda was probably easier than compiling?
18:30:50 <Slereah> Wouldn't that just be replacing the combinator by its lambda?
18:31:00 <ais523> that's what abstraction introduction is
18:31:20 <ehird> but also optimizing
18:32:07 <ais523> here's a program I wrote for the abstraction eliminator:
18:32:08 <ais523> http://pastebin.ca/860594
18:32:18 <Slereah> I wonder, does Unlambda and all the purely functional qualify as string-rewriting?
18:32:37 <ais523> it can be implemented with string-rewriting
18:32:46 <ais523> and that's the method I use when stepping through Unlambda by hand
18:33:05 <ais523> I could probably write a quick interpreter in my new regexp language, which I decided to call Cyclexa
18:34:25 <ais523> the program that I pasted, by the way, is a P'' interpreter in Relambda (which compiles into Unlambda with C-u C-c C-b; it should work without the prefix arg, but it's buggy)
18:34:48 <ais523> Relambda = Unlambda + lambda; any infinite loop must contain a ! character somewhere, which otherwise has the same semantics as i
18:35:11 <ehird> ais523: Does it detect? :-)
18:35:23 <ais523> it detects everything but infinite loops
18:35:31 <ehird> btw, my golf idea: http://golf.shinh.org/, but you can add challenges, and it has a golf challenge for the site source itself
18:35:34 <ais523> which is why you have to put a ! somewhere inside to tell it where they are
18:35:45 <ais523> that's in theory, at least. In practice it seems not to work.
18:35:53 <Slereah> Can't you just solve the halting problem to detect them?
18:37:18 <ais523> how does that golf site work?
18:37:23 <ais523> as in, how do you use it?
18:38:26 <ehird> you see input/output
18:38:30 <ehird> implement the program, etc
18:38:33 <ehird> it goes in the rankings
18:38:45 <ais523> does it check the program to see if it works?
18:39:40 <ais523> and how can Esolang not have a page about GolfScript?
18:39:47 <ehird> yes, it does check
18:39:51 <ehird> http://golf.shinh.org/p.rb?print+file
18:39:55 <ehird> it checks input to output
18:40:03 <ehird> ais523: because its really new
18:40:21 * ais523 must write a stub immediately
18:43:28 <ehird> http://golf.shinh.org/reveal.rb?Quine/shinh/1182045546&php
18:43:30 <ehird> that is so cheating
18:46:32 <ehird> _ is apparently a PHP quine
18:46:39 <ehird> just tested it: yep.
18:46:45 <ais523> as is anything that doesn't contain <?, surely?
18:46:48 <ehird> because it's not in <?php..?>
18:46:53 <ehird> that is so cheating
18:47:07 <ais523> In HOMESPRING, the null program is not a quine.
18:47:23 <ehird> http://golf.shinh.org/p.rb?Quine i'm trying to find an interp with short error messages
18:47:26 <ehird> so ican do a kimian quine
18:47:30 <ehird> I wonder if it has ed
18:48:12 * ais523 had the idea of trying to create an iterating Kimian quine
18:48:26 <ais523> or even better, a quine-pair where one program was Kimian but the other wasn't
18:49:30 <ais523> if I manage to do it in gcc, then that'll be my IOCCC entry for next year
18:49:56 <ais523> my previous one was a BF interpreter in alphanumericless C (apart from defines and the prototype of main)
18:49:59 <ehird> sed: 1: "~": invalid command code ~
18:50:06 <ais523> which contained several hidden programs
18:50:18 <ehird> they use a different sed
18:50:31 <ais523> some of which were written in Perl, some of which were written in BF
18:50:35 <ehird> gnu error messages are large :(
18:51:06 <Slereah> Create a language full of short error messages!
18:51:16 <Slereah> Or some sort of language where the output is only error messages
18:52:56 <ais523> what about this for a Kimian quine, in x86 machine code?
18:53:03 <ais523> Illegal instruction (core dumped)
18:53:25 <ais523> at least on the 8086, all lowercase letters are illegal instructions, so I'd just have to check what capital I does
18:53:55 <ais523> not too long either...
18:54:05 <ehird> "Error: not ELF binary"
18:54:07 <ehird> from the golf site
18:54:17 <ais523> what, they want it in Elf format?
18:54:25 <ais523> just give them your own shorter program
18:54:38 <ehird> it's bsd sed output
18:54:49 <ehird> it outputs it and rejects the submission
18:54:53 <ehird> it checks before verifrying
18:55:06 <ais523> so that's a Kimian quine against the website itself
18:55:12 <ais523> that's a moral victory, at least
18:56:35 <ehird> it checks for status=1
18:56:49 <ehird> I had my D compile time quine:
18:56:50 <ehird> test.d(1): found '1' when expecting ')'
18:56:51 <ehird> test.d(1): no identifier for declarator test.d
18:56:51 <ehird> test.d(1): semicolon expected, not ')'
18:56:51 <ehird> test.d(1): Declaration expected, not ')'
18:57:11 -!- RedDak has quit (Remote closed the connection).
18:57:44 <ehird> it checks if status=1 and denis it
18:58:21 <ehird> test.sh: line 1: [: missing `]'
18:58:30 <ehird> fuck you for matching up yuour []s in errors
18:58:59 <ais523> incidentally, the man page for true gave information on how to make it error out
18:59:19 <ehird> i feel kinda hollow inside
18:59:22 <ais523> /bin/true --version 2>&1 | /dev/full
18:59:37 <ehird> WTF TRUE TAKES OPTIONS
19:00:01 <ais523> only --help and --version
19:00:06 <ais523> and only then if you give a full path
19:00:22 <ais523> and even only then on the GNU version
19:00:34 <ais523> incidentally, /bin/false errors even if you ask for help or version info
19:00:46 <Asztal> because it's so important to know what version of true you're running ;)
19:01:00 <ehird> updated in version 2. now fails.
19:02:00 <ais523> I wonder if anyone ever has run true --help in an attempt to find out what it does?
19:02:30 <Asztal> well, they'd find out alright.
19:03:11 <ehird> http://golf.shinh.org/p.rb?Nothing Um
19:03:38 <ehird> you have to print:
19:03:54 <ehird> with the actual input:
19:04:09 <Slereah> You'd think that most of the shortest programs would be of length zero.
19:04:21 <ehird> you have to print out the string:
19:04:41 <Slereah> Then why are some of the program of length one :o
19:04:54 <ehird> because you get the stirng:
19:05:03 <ehird> you can just do nothing
19:05:06 <ehird> but it denies 0 length programs
19:05:14 <ehird> http://golf.shinh.org/p.rb?Evil+C+Compiler i'm just gonna hardcode this one
19:05:18 <ais523> I just entered qq as a palindromic PHP quine
19:05:31 <ais523> it rejected q; apparently it only wants even-length palindromes
19:05:51 <ehird> wait no... i want to call out to gcc
19:06:43 <ehird> echo"Compiling...\nRunning...";gcc /dev/stdin;./a.out
19:07:13 <ehird> ais523: how do you read form stdin in gcc compiles
19:07:24 <ais523> you have to tell it what language stdin is in
19:08:29 * ais523 is now trying to figure out a palindromic quine in Underload
19:08:39 <ehird> echo 'Compiling...
19:08:39 <ehird> Running...';gcc -xc /dev/stdin;./a.out
19:08:50 <ehird> HAHAHAHHAAHAHAH I RULE
19:09:06 <ais523> I'm sure that's cheating...
19:09:15 <ehird> thats not hte winner
19:09:19 <ehird> ais523: um duh all the submissions are like
19:09:56 <ais523> !daemon ul bf http://pastebin.ca/raw/367774
19:10:36 -!- oklopol has quit (Remote closed the connection).
19:11:49 <ehird> ais523: could i abuse gcc to make it print my lines?
19:14:05 <ehird> ais523: speaking of which
19:14:09 <ehird> how small can a c compiler get?
19:14:24 <ehird> I would venture, 200-300 lines maybe?
19:14:45 <ehird> http://fabrice.bellard.free.fr/otcc/otcc.c otcc is rpetty close, but post-obfuscation..
19:15:04 <ehird> outputting assembly
19:15:08 <ais523> int main(){system("gcc -xc /dev/fd/0");}
19:15:23 <ais523> fd/0 saves one char over your stdin
19:15:30 <ehird> thanks for the suggestion
19:16:24 <ais523> !ul (:aS(:^S^:)Sa:):^S^:(:aS(:^S^:)Sa:)
19:16:25 <ehird> ais523: so what about for a real compiler
19:16:27 <EgoBot> (:aS(:^S^:)Sa:):^S^:(:aS(:^S^:)Sa:)
19:17:56 <ais523> a^ is unlikely to produce output
19:18:05 <ais523> and I can only assume that EgoBot restarts the daemon each time
19:18:17 <ais523> because it's written in BF and does no error checking
19:19:28 <ais523> those are the two main sorts of quines
19:20:15 <ehird> !ul (()*S)(*S)*~:~^
19:20:49 <ehird> ais523: have you got svn on that machine yet? :)
19:20:59 <ais523> two c compilers, in fact
19:21:06 <ais523> there's also a program called c99
19:21:23 <ais523> c99 utility unavailable SunOS 5.9
19:21:52 <ehird> do you have svn yet? :P
19:22:13 <ais523> can't install software on here
19:22:26 <ehird> hahaha, golf.sinh.org pokes fun at ICFP
19:22:28 <ehird> "GolfScript is the programming tool of choice for discriminating golfers.
19:22:28 <ehird> Ruby is a fine programming tool for many courses."
19:22:29 <ais523> it isn't my machine, I'm not root, and I'm not supposed to download any to my home account
19:22:42 <ais523> ICFP is a great idea, by the way
19:23:07 <ais523> oh, and Perl is not too shabby
19:23:14 <ais523> according to their rankings, anyway
19:25:10 <ehird> heh: puts'gXgle'.sub'X','o'*gets.to_i
19:26:17 <fax> 10000000000000000000000000000000000000000000000000000... ?
19:26:51 <ehird> its the 'google' challenge
19:26:57 <ehird> puts"g#{'o'*gets.to_i}gle"
19:27:10 <ehird> the top is 14 chars in golfscript.
19:27:15 <ehird> to beat ruby i need to beat 24 chars
19:28:35 <ehird> i need to optimize gets.to_i
19:30:28 <Asztal> I see String#oct .. but no String#dec :(
19:31:20 <ehird> "42".oct.to_s(8).to_i
19:33:48 <ehird> here's a loop forever program:
19:34:09 <ais523> is that machine code for goto self?
19:34:15 <ais523> or machine code for wait-for-interrupt?
19:34:25 <Asztal> I heard the IOCCC disqualified such things
19:34:41 <ehird> ais523: it works on os x
19:34:47 <ehird> and i dont think golf.shinh.org does
19:34:49 <ais523> that was a rule change after that won the first year
19:34:52 <ehird> i think its portable, in some crazy way
19:34:53 <Asztal> because although compilers would technically accept main not being a function...
19:35:18 <ais523> in C, I think the record is main(){main();}
19:35:30 <ehird> ais523: doesn't work on it
19:35:32 <Asztal> is that valid C? I know it's not valid C++
19:35:32 <ais523> but the compiler needs to do tail-recursion optimization to optimize that
19:35:44 <ais523> C doesn't have a no-recursive-main restriction
19:36:00 <Asztal> in C++, at least, main has undefined linkage and can't be called from inside the program :(
19:36:13 * Slereah is rewriting the Love Machine 9000
19:36:19 <ais523> you can just work around it by making a function that main is a wrapper around
19:36:24 <Slereah> Hoping for some nicer code
19:36:37 <ehird> you can't call main() in C ais523
19:36:50 <ais523> no, you're thinking of C++
19:37:34 * ehird submits ```sii``sii
19:37:48 <ehird> now that's clever.
19:39:04 <ais523> which number is the problem?
19:39:18 <ehird> oh i submitted >< in befunge
19:39:36 <ehird> a one-char befunge
19:40:03 <ehird> ais523: what was it
19:40:04 <ais523> ` leads to a timeout in normal Unlamda because it waits for more input
19:40:08 -!- RedDak has joined.
19:40:13 <ais523> but they remembered to add an EOF
19:40:52 <Asztal> hah, I like this ruby one: `vi`
19:46:14 <ehird> ais523: the only thing better will be mine
19:46:20 <ehird> which as i have asid will have a golf on the site's code itself
19:56:26 <ehird> loooooooooooooooooooooooool
19:56:31 <ehird> ais523: I submitted this to the 123 challenge
19:56:33 <ehird> puts 1;x=File.read("test.rb").sub("puts","puts 1+");File.open("test.rb","w").write(x)
19:56:57 * ais523 looks at the 123 challenge
19:57:58 <ehird> take a look at the top ones
19:58:10 <ais523> I did, it's a clever idea
19:58:15 <ais523> keep resubmitting until it works...
20:00:03 <ais523> what's the sheesh for?
20:00:28 <ehird> because its lame for them to do that
20:02:29 <ehird> puts$_ unless$_=="\n"while gets
20:03:17 <ehird> ,[----------[++++++++++.],]
20:03:25 <ehird> ais523: what's the most compact way to represent 10 in BF?
20:03:34 <ais523> there's a page about it on the wiki
20:03:55 <fax> +++++[>++<-]>
20:03:57 <ais523> and yes, all numbers up to 14 are best done in unary
20:04:04 <ehird> ais523: but i have duplication there
20:04:57 <ais523> incidentally, I like my Perl entry for the echo, even though it's clearly not the shortest
20:05:23 <ais523> the whole program is preprocessor directives !
20:06:30 <ais523> most people probably don't even realise that you can do that
20:07:42 <ais523> likewise, a perl quine:
20:08:01 <ais523> I don't even know if that's cheating or not
20:08:13 <ehird> also, it's an X quine for all X
20:08:39 <ais523> no, because only perl will run a different program with the current source if it sees a #! at the start
20:08:44 <ais523> cat doesn't, for instance
20:09:12 <ehird> print gets.split('').sort
20:09:16 <ehird> http://golf.shinh.org/p.rb?sort+characters
20:09:19 -!- oerjan has joined.
20:11:04 <ehird> we were discussing golf.
20:11:39 <oerjan> one of the many sports i have not tried.
20:11:39 <ais523> I went and entered a new competition: to create the shortest Underload interpreter
20:11:57 <oerjan> unless you count minigolf.
20:12:08 <ais523> this is the programming kind of Golf
20:12:24 <ehird> ais523: Our compiler won't do very well. :-)
20:12:36 <ais523> I allowed execs, though, so it will be allowed...
20:13:14 <ehird> why did you allow execs?
20:13:26 <ais523> so that you can pipe a compiled output into gcc
20:13:39 <ais523> and because I couldn't see a reason to deny them
20:14:06 <ehird> 'd need to include prelude.c
20:14:13 <ehird> and that would make it biiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiig ;)
20:14:33 <ehird> there should be an anti-golf
20:14:43 <ehird> 'find the most obscure, drawn out, just plain ridiculous solution'
20:16:02 <Asztal> all the COBOL users would win on file size :(
20:17:21 <Slereah> Well, you could probably make some program that will send an email to a third party, asking him to send a letter containing "Hello world!"
20:17:46 <Slereah> Ideally, on another continent
20:17:51 <ehird> which is then OCR'd in
20:18:07 <Asztal> by writing an x86 emulator, an OS, with a TCP/IP stack and sendmail
20:19:31 <Slereah> Are there any company that will send snailmails containing some text automatically?
20:19:42 <Slereah> Relying on a third party feels a little like cheating.
20:22:15 <RodgerTheGreat> you'd still have to write a "driver" for the third party. Think of it as a piece of hardware.
20:25:58 <ehird> Slereah: man: a very fancy machine
20:26:03 <ehird> for some values of 'very fancy'
20:27:05 <Slereah> Well, depends in what areas.
20:27:18 <Slereah> We can walk without bumping into walls!
20:27:32 <Slereah> You can't say that of every robot
20:30:47 <Slereah> Plus, we're not that slow!
20:30:56 <Slereah> It just takes us a while to speed up.
20:31:06 <Slereah> A few thousand years, the time for us to invent computers.
20:32:00 -!- puzzlet_ has quit (Remote closed the connection).
20:32:07 -!- puzzlet has joined.
20:32:55 <RodgerTheGreat> plus our collective processing power is capable of growing at a nearly exponential rate
20:33:36 <ehird> so! who wants to write a tiny esoteric C compiler
20:33:44 <ehird> aimed to be the smallest self-compiling C compiler
20:33:48 <ehird> tiny-esoteric-subset-of-C
20:34:40 * Asztal foresees abuse of goto
20:34:48 <RodgerTheGreat> sounds vaguely interesting. What platform will we target?
20:35:14 * fax wants an ISWYM impl.
20:35:22 <RodgerTheGreat> let's make goto capable of calling function pointers or something
20:35:26 <oerjan> wasn't that done already as an IOCCC winning entry?
20:35:37 <ehird> oerjan: maybe, but whatever
20:35:42 <ehird> RodgerTheGreat: i was thinking just portable asm
20:35:50 <ehird> 1. it's easier to debug :)
20:35:59 <ehird> i use OS X, you probably use something else.
20:36:10 <ehird> well someone else might not
20:36:14 <fax> what about replacing goto with the J operator (not that I actually know what it does)
20:36:27 <RodgerTheGreat> lies and heresy. Everyone uses OSX or a degenerate form of OSX.
20:36:30 <ehird> fax: that's a bit too changed from C
20:36:34 <ehird> hopefully we can make the only thing requiring special stuff syscalls
20:36:48 <RodgerTheGreat> It's very much like how all labor is really degenerate forms of bending.
20:36:51 <ehird> also, this code should compile OK:
20:37:00 <ehird> int main(void) { printf("Hello, world!\n"); return 42; }
20:37:02 <ehird> when we're done obviously.
20:37:07 <ehird> int main(int argc, char **argv) { printf("Hello, world!\n"); return 42; }
20:37:11 <ehird> so: we do need types
20:37:19 <ehird> RodgerTheGreat: let's not :P
20:37:29 <ehird> bit hard to jump right into c like that
20:37:56 <ehird> RodgerTheGreat: Oh yeah and no parser generators
20:37:57 <RodgerTheGreat> will we also have to build portions of a standard library from scratch?
20:38:04 <ehird> and... maybe. maybe not. we'll see
20:38:09 * ais523 did a 303-byte Underload interpreter in Perl
20:38:13 <RodgerTheGreat> ehird: I never use parser generators. They're for pussies and people in a rush
20:38:21 <ehird> RodgerTheGreat: though i don't think we should compile to _main
20:38:28 <ehird> because the C stdlib does that stuff
20:38:33 <ehird> and well it just seems like too much of a copout :P
20:38:36 <ais523> it's actually an optimising interpreter, so it's not quite correct because it sometimes optimises inside literal strings, but the examples don't catach that problem
20:38:40 <ehird> but... entry points are wildly different between platforms
20:38:56 <RodgerTheGreat> I'm just trying to firmly establish what we plan to accomplish and create from scratch
20:39:34 <ehird> RodgerTheGreat: yeah
20:39:51 <ehird> RodgerTheGreat: OK i guess we can compile to just main for now
20:39:55 <fax> make 'int' arbitrary precision
20:39:57 <ehird> and let the c runtime do the legwork.
20:40:33 <ehird> RodgerTheGreat: do we want to typecheck? :D
20:41:02 <RodgerTheGreat> it'll make it simpler, and we can just assume C programmers know what they're doing
20:41:12 <Asztal> everything is an int! including pointers!
20:41:16 <ehird> but that's not as fun RodgerTheGreat :)
20:41:31 <RodgerTheGreat> chars will actually be ints, booleans will actually be ints, etc
20:41:41 <fax> everythin is arbitrary precision int
20:41:47 <RodgerTheGreat> we can encode things differently, but you can randomly swap however you want
20:41:47 <ehird> RodgerTheGreat: but imagine golfing a typechecker
20:41:53 <ehird> we could add type inference
20:42:19 <ehird> it's not C without type checking
20:42:39 <RodgerTheGreat> ideally with as little undefined behavior as is reasonably possible
20:42:51 <ehird> But we're trying to compile something that is recognizable as C here :P
20:43:55 <RodgerTheGreat> we're trying to make something that is A) a programming language, B) is compatible with C functions/libs/etc, and C) can compile the test programs "int main(void) { printf("Hello, world!\n"); return 42; }" and "int main(int argc, char **argv) { printf("Hello, world!\n"); return 42; }" like you'd expect them to in regular C, without cheating.
20:44:19 <ehird> you are splitting hairs here, though
20:44:54 <ehird> why can't we just aim to write "A c compiler" and then decide what to not implement when we come to that?
20:45:06 <ehird> But type checking is very fundamental to *C*, one of its defining things, so..
20:45:13 <RodgerTheGreat> I am reducing the scope of the problem until we have solidified a design. This is generally how I work.
20:45:27 <ehird> yes, but this way we'll be going through the whole spec
20:46:12 <RodgerTheGreat> if you don't nitpick, how can you collaboratively design something as necessarily precise as a programming language?
20:46:15 <Asztal> what about the preprocessor?
20:46:37 <ehird> RodgerTheGreat: we're not designing a programming language. we're implementing a language, continually reducing it.
20:46:42 <ehird> it's incremental development.
20:46:47 <ehird> Asztal: i think that can come last.
20:46:53 <RodgerTheGreat> Asztal: it seems reasonable that we could do without a preprocessor, unless it allows us to get away with a simpler language
20:47:15 <ehird> a preprocessor will only be a few hundred lines anyway
20:47:23 <ehird> we're going for 'conciseness', not 'bytes'
20:47:30 <fax> A nice macro system would be cool
20:47:53 <fax> concise CLISP
20:48:01 <RodgerTheGreat> a nice macro system and a powerful goto might make it possible to have entirely synthetic functions! XD
20:48:16 <ehird> macros with return types!
20:48:23 <ehird> they define the type of the expression they produce
20:48:52 <RodgerTheGreat> all we need to do is come up with a way for macros to generate something a linker can work with
20:49:02 <ehird> RodgerTheGreat: Macros are easy:
20:49:12 <ehird> docompile(macro_body);
20:49:16 <ehird> call_assembler(that);
20:49:29 <ehird> call_linker(that,"blah");
20:49:35 <ehird> ... get a funcptr ...
20:49:40 <ehird> funcptr(argstomacro);
20:49:56 <ehird> RodgerTheGreat: you can just grab a funcptr from the linked file
20:50:28 <RodgerTheGreat> sounds very creative, and it will significantly reduce the complexity of the language, I think
20:50:42 <ehird> nah, i don't think so
20:50:45 <ehird> it'll be a fun extra
20:50:51 <RodgerTheGreat> recursion may cease to be possible, depending on how this works, but that's fine.
20:50:54 <ehird> but i don't think it'll provide much actual usefulness or ways to reduce
20:51:39 <ehird> RodgerTheGreat: by the way, what source control systems do you have?
20:51:45 <ehird> pleasesaymercurialordarcspleasesaymercurialordarcs
20:52:30 <ehird> are you suggesting that we just ftp it somewhere and overwrite everything?
20:52:35 <ehird> darcs and hg aren't like svn and the like
20:52:58 <ehird> so you have experience with subversion and didn't like it
20:53:09 <ehird> you should try a distributed system like darcs or hg though.
20:53:24 <RodgerTheGreat> subsubversion solves most of my objections to source-control systems in general
20:56:01 * sekhmet can't stand *not* using VCSes
20:56:55 <ehird> anybody want to work on the compiler that doesn't despise vcses? :P
20:56:58 -!- sebbu has joined.
21:01:39 <ehird> RodgerTheGreat: so are you adamant about this vcs thing? :P
21:02:08 <ehird> not even going to *try* darcs or hg?
21:02:14 <ehird> which are nothing LIKE svn?
21:03:23 <RodgerTheGreat> there is a sizeable portion of my soul dedicated to my hatred of version control systems. If I am getting paid, I will use one, but I at one point stated that I would never use one for personal projects
21:03:51 <sekhmet> What do you hate about them, out of curiosity?
21:04:08 <ais523> just for future reference:
21:04:09 <ehird> RodgerTheGreat: you really think they're all like svn don't you..
21:04:10 <ais523> '('/'{'*')'/'}'*'S'/'p'*'a'/'`'*'~'/"\\"*'!'/';'*'*'/'+'*':'/'.'*'^'/'~'*
21:04:20 <ais523> is almost a Underload -> GolfScript compiler in GolfScript
21:04:29 <ais523> except that the output is completely wrong
21:04:38 <ais523> because it translates string literals into GolfScript too
21:04:40 <ehird> ais523: make it store the original string.
21:04:50 <ehird> ais523: then add an 'eval' and submit it to your competition :P
21:04:53 <ais523> not very easy, I don't think
21:05:04 <RodgerTheGreat> they seem to me an inherently flawed design. Merging, forking and the like can be resolved with policy, rather than software. They also almost universally force people to install the program just to check out the latest ticket, which is stupid for many reasons.
21:05:11 <ais523> because it simply literally converts equivalent commands
21:05:22 <ehird> RodgerTheGreat: Yes, let's do what the computer can do for us!
21:05:33 <ais523> both languages are concatenative, and in such cases a one-to-one conversion is often possible
21:05:35 <ehird> And you can check out svn, darcs, ... just with a recursive wget.
21:06:22 <ehird> oh i'm sorry i forgot that typing one line is impossible
21:06:27 <ehird> let's just click instead :|
21:07:03 <RodgerTheGreat> there are many reasons I use a mac, and one of them is that I have no patience whatsoever for a shitty UI, be it console-based or graphical.
21:07:17 <ehird> ... wget is a shitty UI?
21:07:32 <ehird> wget is a shitty ui?
21:11:18 <ais523> so, is this bash quine a quine by cheating?
21:11:36 <ehird> ps: %a: keyword not found
21:11:37 <ehird> ps: no valid keywords
21:11:38 -!- sebbu2 has quit (Connection timed out).
21:11:50 <ais523> it looks to me like on some system, it works by ps'ing to find its own command line
21:12:59 <ais523> wow, someone's come up with a 76-byte Underload interpreter in C
21:13:04 <ais523> I wonder if it's for real or by cheating?
21:13:31 <ais523> has to be, really, doesn't it
21:13:45 <ais523> maybe it looks at the input and comes up with one of the three possible outputs depending on what the input was
21:13:50 <Slereah> But, is it good cheating or bad cheating?
21:14:07 <ehird> ps -Aocommand|grep "ps -"
21:14:10 <ehird> ^ working ps quine
21:14:29 <ehird> ais523: and yeah of course
21:14:46 <ais523> not until the deadline
21:14:49 <ais523> which I set in 4 weeks time
21:14:56 <ehird> ais523: bah, 4 weeks?
21:15:04 <ais523> may as well leave it up there a while
21:15:15 <ehird> 51b is a big user of that site
21:15:15 <ais523> so that's a clue that it's a cheat
21:15:17 <ehird> so.. it's possible
21:15:31 <ehird> but what's (embed)
21:15:41 <ais523> my guess is that the answers are embedded in the source
21:15:58 <ehird> then 51b pointed it out
21:18:43 <ais523> I did it probably the same way in 37 by cheating
21:19:49 <ais523> just switch between the 3 possible answers based on whether certain characters are in the source
21:20:09 <ehird> ais523: does an underload interp reeaallly need an infinite stack
21:20:14 <ehird> or is, say, 65536 enough
21:20:41 <ais523> according to the spec, it needs an infinite stack, but it also has to be capable of infinitely long strings, etc., regardless of your computer's memory
21:20:47 <ais523> in real life you can put limits on it
21:21:22 <ehird> yeah i am going to for speed
21:21:27 <ehird> ais523: what about the length of the strings?
21:21:37 <ehird> i.e. (..morethan1024chars..)=err
21:21:42 <ais523> not nearly, because numbers are generally stored in unary
21:21:47 <ais523> taking up two characters per value
21:22:07 <ais523> 65540 is probably enough
21:22:18 <ais523> that lets you store numbers up to 32767 with a bit of leeway for doing arithmetic
21:22:50 <ais523> why do you want a string length limit anyway?
21:22:55 <ais523> Are you trying to avoid malloc?
21:23:09 <ehird> ais523: no -- trying to avoid realloc()
21:23:17 <ehird> cause you have to store the length, check, etc
21:23:52 <ais523> in that case, make the limits just long enough to run the example programs
21:24:11 <ehird> i wish you could get the length of stdin in c
21:24:31 <ais523> putting a limit on the /length/ of an input program is probably acceptable
21:24:40 <ehird> i'm just doing it in ruby
21:24:42 <ehird> far easier to golf
21:24:50 -!- Corun has joined.
21:26:35 <ehird> ais523: i'm only handling one line programs
21:26:50 <ais523> it isn't as if newlines are legal outside strings anyway
21:31:39 <ehird> huh. the compiler allows that.
21:31:45 <ehird> in fact it allows all junk
21:32:11 <ais523> it uses the INTERCAL rule; junk only causes an error if it's executed
21:32:22 <ais523> so just compile an error function as the code and use the correct string
21:33:41 <ais523> you just need to map all unrecognized characters to call error() just like ~ is mapped to swap(), etc.
21:35:30 <ais523> not slow, but inefficient
21:35:45 <ais523> hmm... maybe you could not generate any code in a block after the first error()
21:35:54 <ais523> after all, nothing past that point can be run anyway
21:38:08 <ehird> ais523: damn mine is 325
21:38:19 <ais523> what sort of method did you do?
21:38:20 <ehird> very helpful ruby.
21:38:33 <ehird> s=(gets.split"").map{|x|x[0]}.reverse;
21:38:37 <ehird> ais523: ^^ probably what takes the space
21:39:39 <ais523> it doesn't evaluate in the right order - it's a strange mix of lazy and eager - but that doesn't matter for the example problems because it only affects the S command
21:40:51 <ehird> % echo "(:aSS):aSS" | ruby ul.rb && echo
21:41:17 <ehird> % echo "(x)(::**)(::**)^^S" | ruby ul.rb && echo
21:41:17 <ehird> ul.rb:1:in `[]=': index -2 out of array (IndexError)
21:41:54 <ehird> % echo "(x)(::**)(::**)^^S" | ruby ul.rb && echo
21:41:54 <ehird> x::**::**x::**::**x::**::**x::*............
21:42:04 <ehird> a=[];s=(gets.split"").map{|x|x[0]}.reverse;while s!=[];c=s.pop;case c;when ?(;i=1;a<<[];while i>0;x=s.pop;if x==?(;i+=1;elsif x==?);i-=1;else a[-1]<<x;end;end;when ?:;a<<a[-1];when ?!;a.pop;when ?^;s+=a[-1];when ?~;x=a[-1];a[-1]=a[-2];a[-2]=x;when ?*;x=a.pop;a[-1]+=x;when ?S;print a.pop.map{|x|x.chr}.join("");when ?a;a[-1]=[?(]+a[-1]+[?)]end;end
21:42:09 <ehird> debug my code for me! :-)
21:42:42 <ais523> the error doesn't spring out at me
21:43:11 <ehird> ais523: im gonna write one in sed..
21:43:13 <ais523> does ^ remove an entry from the stack?
21:43:40 <ais523> I might be able to get a sed version by translating from the Thutu, but I wouldn't really want to try
21:43:54 <ehird> ul.rb:1: undefined method `+' for nil:NilClass (NoMethodError)
21:44:06 <ehird> ais523: Give me a good code/stack seperator
21:44:18 <ais523> because it's banned in the original language
21:44:20 <ehird> any single char not an underload command
21:44:28 <ehird> then seperate with.. what?
21:44:32 <ehird> that's what your js does
21:44:45 <ais523> yup, I was preparing to retrofit the " syntax in later
21:44:50 <ais523> but it never really caught on
21:45:05 <ais523> so it remains part of the spec but never implemented, and it's probably better to keep it that way
21:45:19 <ais523> besides, an escape character is useless if it can't even escape parens
21:45:31 <ais523> it only escapes <>[]", which are only banned because I say so
21:45:59 <ehird> My basic stuff is going to be: s/INSTRUCTION([^<]*)(([^<]<)*)TOP ELEMENT/RESULT/g
21:46:16 <ehird> paren matching will be too hard
21:48:11 <ais523> the Thutu method is to change inner parens to {} temporarily
21:48:25 <ais523> but that makes up a good part of the program
21:49:22 <ehird> i submitted a cheat one.
21:49:23 <ehird> s/.*^^S/xxxxxxxxxxxxxxxxxxxxxxxxxxx/
21:49:23 <ehird> s/.*^S$/::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::/
21:49:34 <ehird> but forgot to say cheat
21:49:56 <ehird> ais523: Write an UL interp in your regexp lang
21:54:22 <ais523> (```s$+$+$+!=``=1=3`=2=3|``k$+$+!==2|`i$+!==1|`$+$+|[ski])=^*?('ski'|`'sk'$+|``s$+$+)$
21:54:28 <ais523> does it for ski unlambda
21:54:40 <ais523> but it'll go into an infinite loop on an invalid program
21:55:08 <ais523> oh, as for forgetting to say cheat, resumbit it as ehird(otherthesamelengthisacheatandsoisthisone)
21:57:17 <ehird> http://golf.shinh.org/p.rb?Minimal+scheme+interpreter
21:57:21 <ehird> possibly the hardest one yet :P
21:57:46 <ehird> barely golfable actually
21:58:34 <ehird> ais523: if you even get a submission that works with that in less than an awful lot of code i'll be impressed :P
21:59:23 <ais523> pretty cheatable, of course
21:59:39 <ehird> and i missed out things like (quote X)
21:59:43 <ais523> does Scheme have an eval?
21:59:58 <ehird> you don't need to provide it though
22:00:04 <ais523> no, but you could use it
22:01:23 <ehird> (let l()(let((_(read)))(if(not(eof-object? _))(eval _))(l)))
22:01:56 <ehird> have to specify an environment
22:02:31 <ehird> i have unbalanced parens in my test
22:03:08 <ehird> lfjklsdhfsdjklfhs WHY CAN'T YOU EDIT IT
22:04:12 <ais523> submit a 'fixed version' of the problem, and mention in the description that the other has typos in the test cases
22:04:51 <ehird> ais523: i'm just submitting a lambda calculus one
22:04:54 <ehird> with unlambda-lambda-syntax
22:04:59 <ehird> but with multichar identifiers
22:05:03 <ais523> abstraction elimination?
22:06:05 <ehird> ais523: (a b) in lambda calculus
22:06:17 <ehird> ais523: should i specify regular lambda calculus syntax? :)
22:06:17 <ais523> so you use parens for disambiguation?
22:06:33 <ais523> make it Unlambda syntax, just with lambdas
22:06:47 <ehird> i think i should make it real lambda calculus
22:08:24 <fax> Why does no one use ??
22:08:40 <ais523> I use ?? in Perl a lot
22:09:16 <fax> I meant lambda character
22:10:08 * ehird makes golfscript-inspired language
22:10:21 <ehird> everything is a stack: even the program
22:11:15 <ehird> { a b c } -> stack with top=a, prev=(top=b, prev=(top=c, prev=()))
22:11:19 <ehird> and you can call it
22:11:25 <Slereah> Around stacks, never relax.
22:11:35 <ehird> the whole program is implicitly in {...}!
22:12:58 <ehird> and the main stack is accessed as $
22:13:32 <ehird> STACK={ }, PROG={ { 1 } { 1 + } $ ~ ! }
22:13:39 <ehird> STACK={ { 1 } }, PROG={ { 1 + } $ ~ ! }
22:13:45 <ehird> STACK={ { 1 } { 1 + } }, PROG={ $ ~ ! }
22:13:56 <ehird> STACK={ { 1 } { 1 + } <STACKPTR> }, PROG={ ~ ! }
22:14:10 <ehird> STACK={ { 1 1 + } }, PROG={ ! }
22:14:31 <ehird> >>STACK={ }, PROG={ 1 1 + }
22:14:39 <ehird> >>STACK={ 1 }, PROG={ 1 + }
22:14:43 <ehird> >>STACK={ 1 1 }, PROG={ + }
22:14:47 <ehird> >>STACK={ 2 }, PROG={ }
22:14:56 <ehird> STACK={ 2 }, PROG={ }
22:15:09 <ais523> my stack trick for the genuine Perl Underload interpreter was to just leave (...) constructs at the left-hand end of the program as a stack, and to execute the first real command after them
22:15:18 <ais523> saves having to implement (...) at all
22:16:30 <ehird> someone step through:
22:16:32 <ehird> {0={1}{:1-#!*}?}4!
22:17:26 <ais523> not very easy when we don't know the syntax
22:17:41 <ehird> you got a kind of feel for it up there
22:18:59 <ais523> maybe 2008 will be the year of concatenative languages?
22:19:24 <ais523> I've spent a while thinking about one which also has arrays, pointers and GOTO
22:19:33 <ais523> so as to be able to use as many paradigms as possible
22:23:15 -!- ais523 has quit ("wow, it's late").
22:24:21 <ehird> so, what languages should i allow in my golf?
22:27:21 -!- RedDak has quit (Remote closed the connection).
22:29:11 -!- oerjan has quit ("Good night").
23:04:11 -!- cherez has quit (Read error: 110 (Connection timed out)).
23:09:42 -!- helios24 has quit ("Leaving").
23:14:13 -!- jix has quit (Read error: 113 (No route to host)).
23:16:36 -!- jix has joined.
23:23:30 -!- immibis has joined.
23:45:42 -!- cherez has joined.
23:49:03 <EgoBot> help ps kill i eof flush show ls bf_txtgen usertrig daemon undaemon
23:49:05 <EgoBot> 1l 2l adjust axo bch bf{8,[16],32,64} funge93 fyb fybs glass glypho kipple lambda lazyk linguine malbolge pbrain qbf rail rhotor sadol sceql trigger udage01 unlambda whirl
23:52:59 <immibis> ehird: obviously everyone in the channel is alive (or most people anyway)
23:53:06 <ehird> it was meant figuratively
23:53:08 <immibis> well it depends. what is life?
23:56:08 -!- Corun has quit (Read error: 110 (Connection timed out)).
23:56:54 -!- timotiis has quit ("leaving").
23:57:06 -!- timotiis has joined.