freenode/#lisp - IRC Chatlog

5:49:35 kenanb maybe it is set to etags search function or something, which might be causing the definitions to not be found. -rpc version asks swank to find it, which works successfully in my system

5:50:04 Bike ugh, of course it's not etags. it's rpc, but it doesn't work for some sbcl internals lately, for me, that's all.

5:51:50 kenanb hmm weird. some emacs plugins try to be extra smart with behaviour of irrelevant libraries, I hoped it might be the case that some library is messing with the search function

5:52:56 Bike i'm not very good at emacs. the most complicated thing i use is slime and my rc is like ten lines. so it's just some voodoo wizard fucking with me.

5:54:49 kenanb it only partially doesn't work?

5:55:00 kenanb very weird indeed

5:56:10 Bike oh, even better, if I enter sb-c::%defknown myself it does work, just not when doing it from a buffer. oh well, it's not a big deal.

5:58:24 kenanb you mean when you enter it yourself interactively to slime-edit-definition from minibuffer?

5:58:55 Bike yes

5:59:17 Bike so presumably there is some failure to interrogate the correct symbol from the file buffer

5:59:54 kenanb maybe slime cannot properly determine the package at point

6:00:45 kenanb can you try after prefixing the package designator to a function in a buffer

6:01:27 kenanb tho if it cannot determine the package at point, it shouldn't be able to eval region etc as well,

6:01:40 Bike oh, wait, i know the problem, it's sbcl's ! thing. no big.

6:01:57 Bike probably.

6:03:24 kenanb ah, it might be, tho I wonder how my slime can find if that is the case.

6:03:39 kenanb but still seems like a usual suspect

6:05:34 kenanb Bike: maybe this? http://askubuntu.com/questions/392118/slime-lookup-definition-not-working-with-sbcl

6:06:11 kenanb tho then it probably wouldn't be able to find when you enter it interactively

6:06:49 PuercoPop anyone know if parenscript used to export true symbol? if so what happened to it?

6:19:24 itheos ** NICK theos

6:22:35 shka hi

6:22:44 shka what's the difference between struct and class?

6:23:00 shka and does integer slots in struct are boxed?

6:23:50 beach shka: You seem to be a bit confused.

6:24:01 shka i am

6:24:04 shka that's true

6:24:16 beach shka: Every Common Lisp object is an instance of some class.

6:24:20 shka but can't seem to find good explanation in common lisp the language

6:24:31 shka ok, i know that

6:24:57 kenanb shka: then as you can guess, a struct is a class

6:25:13 beach Then, it depends on what you mean by "struct". Do you mean the TYPE created by DEFSTRUCT, or do you mean the instance of that type?

6:25:14 shka but still, i'm under impression that defstruct and defclass construct object in a different way

6:25:29 shka beach: i mean instance

6:25:45 shka well, defstruct just declares

6:25:55 shka but hopefully you know what i mean here

6:26:03 beach So you are comparing an instance of DEFTYPE to a CLASS created by DEFCLASS?

6:26:08 beach That's ... special.

6:26:17 shka no

6:26:23 beach That's what you just said.

6:26:28 shka oh

6:26:41 shka let me rephrasse that

6:27:07 shka i want to know the difference between object of a class declared by defstruct with object of a class declared by defclass

6:27:21 beach Ah, that's different.

6:27:29 shka sorry for confusion

6:27:38 Bike structs are more likely to be boxed. classes have so much redefinition stuff it's hard to allow things.

6:28:02 beach Bike: What?

6:28:19 beach "unboxed"?

6:29:18 Bike more likely to have unboxed slots, i meant. yes.

6:29:40 shka ok

6:29:40 kenanb shka: tho the standard doesn't mandate such optimizations, it is up to implementation to do such things

6:30:17 shka kenanb: it's ok as long ccl and sbcl do that

6:30:19 shka at least for me

6:30:36 shka ok

6:30:40 shka what about readers

6:31:04 Bike what about them?

6:31:33 shka why defstruct declare readers with struct name prefix?

6:31:46 Bike because it seemed like a good idea at the time.

6:31:53 Bike you can change the behavior a bit with the :conc-name option.

6:31:56 shka oh

6:32:49 shka so i can have something like access-x for two classes declared by defstruct?

6:33:48 Bike what?

6:34:22 shka eeeeh i can actually check that myself

6:34:33 kenanb shka: you may think of a structure object as kind of like a very primitive alternative to a CLOS standard object. they are likely to be optimized more heavily due to being less powerful than a standard object, they don't support multiple inheritance, the slot access doesn't necessarily follow the same strict rules as the ones for a standard-object. so more implementation dependent behaviour. what you usually have with a structure is the

6:34:33 kenanb ability to change the interface, rather than the behaviour, unlike CLOS in general

6:34:37 shka ok, thanks guys for explanation

6:36:01 shka kenanb: what i need is basicly compact data record

6:36:17 shka i don't want to even do dynamic dispatch on it

6:38:41 shka Bike, beach, kenanb thanks you all guys

6:38:44 kenanb shka: then they might be useful, careful initializing your slots. they will be initialized to "something" if you don't provide initforms.

6:39:08 shka kenanb: yeah, i figured that out

6:40:25 kenanb shka: or they may not be initialized by default, nothing seems to be guaranteed on that end really. always provide initforms.

6:41:18 kenanb shka: I would just go with standard objects unless I am sure I want structs for some reason

6:41:37 shka kenanb: i'm sure

6:44:13 kenanb shka: structures might be of represented as vectors or lists by explicitly specifying a :type, in which case there will really be no new class definition, therefor no method to specialize on it, therefore all accessors will really be functions, so it is highly likely that you will accidentally have name clashes because of automatically generated reader names, with that in mind, prefixing struct name seems like a good idea, just to make

6:44:13 kenanb sure.

6:44:36 shka kenanb: gotcha

6:44:40 shka i noticed that

6:45:26 shka i was expecting that new type would be generated even for :type vector

6:45:32 shka but that's not the case

6:45:40 Zhivago Remember that types and classes differ.

6:45:48 shka yeah

6:45:56 shka i'm keeping this in mind

6:46:00 shka but that's what happend

6:46:09 shka i wrote function with input

6:46:15 shka i declared type of the input

6:46:31 shka and sorry, but my argument is of type vector

6:46:56 shka after a moment i noticed that basicly no additional type was declared

6:46:59 Zhivago You can always wrap it in a little class entity.

6:47:18 shka it's ok, i really don't need it

6:47:26 shka i

6:47:34 shka i'm ok as long as my slots are not boxed

6:47:46 shka in just one class

6:48:47 kenanb shka: an inferred-type can be generated if you also provide :named option. because there will be a struct-name-p which will simply check the symbol naming the struct in corresponding position of vector/list. if you don't provide it, there is no way to retrive that information, it is just a sequence. but that can only be used to generate an inferred type (like in deftype), not a class representing the structure.

6:49:11 shka aha!

6:49:23 shka kenanb: thanks, i would never guess that

6:49:26 Zhivago And there's always satisfies. :)

6:50:03 kenanb shka: well, I would never guess that, too. thankfully, standard states that very clearly. :D

6:55:17 kenanb Zhivago: that was not a happy smile, was it?

6:55:41 kenanb I feel strong hate towards satisfies :)

6:56:27 jackdaniel you are not satisfied with it? ;)

6:57:47 shka why?

6:58:05 shka i like satisfies

6:58:13 shka jackdaniel: what a dire pun, sir

6:58:30 kenanb :) not me tho, I am not knowledgable enough on types to even consider having feelings about a particular type specifier, tho I know I simply ignore satisfies clauses in the type expansion code I am currently writing

7:27:16 kenanb Bike: worked like a charm!! pushed fix, created pull request. https://github.com/froydnj/nibbles/pull/22

7:28:43 kenanb tho travis-ci test failed, weird

7:30:49 kenanb silly me, fixing

7:56:18 kenanb oh my god explosion in Brussels Airport :(

7:57:08 |3b|`` ** NICK |3b|

8:09:46 aap_ ** NICK aap

8:30:44 myrkraverk When I have a hashtable where the values are lists, how do I append to such a list?

8:32:28 myrkraverk I can't seem to get it right; in particular (and maybe predictably, I don't know) this doesn't seem to work: (setf (gethash normalized-word words) (append (gethash normalized-word words) (list count))))

8:33:23 Munksgaard myrkraverk: You probably need to use nconc instead of append.

8:33:55 Munksgaard Append returns a new function (is "pure"), while nconc modifies the lists in-place.

8:34:20 Munksgaard s/function/list/

8:34:51 myrkraverk Hmm, ok.

8:34:56 kenanb no

8:35:30 myrkraverk no? nconc is not the way?

8:35:46 kenanb no, that is pretty irrelevant

8:36:16 kenanb explicitly setting something should work

8:37:04 kenanb can you paste the whole code

8:37:14 myrkraverk Ok, then (maybe) my comparison function in the hash table is wrong. IIRC gigamonkeys said to use 'equal for string keys, so I did that.

8:37:24 kenanb I mean that snippet is part of a function I assume?

8:38:03 kenanb normalized-word is supposed to be a parameter I guess, since it is not quoted

8:38:16 myrkraverk More like a loop at this point, but I'll try to come up with a minimal self contained example. And it just might be the case that my minimal example will work, in which case: back to the drawing board.

8:38:24 myrkraverk Yeah, it's a parameter.

8:38:46 kenanb are you looping in hash table?

8:39:12 Munksgaard kenanb: I guess you're right.

8:42:11 myrkraverk Do, I just realize my stupidity; the hash table is defined in the wrong loop (inner, not outer)

8:42:15 myrkraverk *Doh

8:42:25 myrkraverk Sorry about the mess.

8:43:08 kenanb no worries it happens, good luck

8:52:14 malice Hello, I want to print minutes with format; I want to print the number as with ~D, except when it's less than 10 - then I want to add 0 before the number

8:52:35 malice e.g. for numbers: 4=>04 ; 15=>15; 0 =>00

8:52:44 malice What would be the best way in format to do this?

8:55:55 kenanb malice: (loop for i to 20 do (format t "~2,'0d " i)) test this

8:58:30 malice kenanb: Works, nice! Thanks

10:52:44 John[Lisbeth] Ok I am reading the gigamonkeys thing and I am inside slime. And I've typed (defun hello-world () (format t "Hello, World!"))

10:52:56 John[Lisbeth] And the output I got was HELLO-WORLD

10:53:07 John[Lisbeth] so I am wondering what that output means and what type it is

10:54:16 djh It's the function name

10:54:35 djh You've defined the HELLO-WORLD function

10:54:40 ecraven John[Lisbeth]: defun in this case probably 'returns' the symbol of the function name

10:54:52 John[Lisbeth] So can I pass that into another function and have it be the same as hello-world ?

10:54:56 p_l John[Lisbeth]: you have defined a function, called HELLO-WORLD, which is what defun returned

10:55:19 p_l John[Lisbeth]: you can now call it with (hello-world) - or with (HELLO-WORLD)

10:55:30 djh You can pass it in via #'HELLO-WORLD

10:55:32 ecraven (HeLlO-wOrLd) :p

10:55:33 djh as well

10:55:35 John[Lisbeth] What I am saying is

10:56:18 John[Lisbeth] (== (HELLO-WORLD) (defun hello-world () (format t "Hello, World!")))

10:56:25 codesine (setf *foo* (defun hello-world ()))

10:56:29 codesine then do

10:56:33 codesine (type-of *foo*)

10:56:43 codesine the HELLO-WORLD after the defun is the symbol of the function returned

10:57:30 codesine (equal *foo* 'HELLO-WORLD)

10:58:33 codesine in your case

10:58:34 John[Lisbeth] I can't get that first line to compile

10:58:44 codesine (equal 'HELLO-WORLD (defun hello-world()))

10:59:00 John[Lisbeth] I can't compile your example of defun

10:59:38 codesine (equal 'HELLO-WORLD (defun hello-world ()))

10:59:38 John[Lisbeth] Ok last examples works

11:00:05 John[Lisbeth] Alright so they are equivelant then

11:00:22 John[Lisbeth] That is what I need to know, ty

11:00:32 codesine the return value is just the<codesine> (equal 'HELLO-WORLD (defun hello-world ()))

11:00:32 codesine <John[Lisbeth]> Ok last examples works

11:00:32 codesine ERC

11:00:45 codesine cool, that book is good

11:01:26 John[Lisbeth] I was doing sicp but then I realized I didn't want to do things purefly fuctionally

11:01:31 John[Lisbeth] and before that I was doing haskell

11:01:44 John[Lisbeth] and before that bash, then C++

11:01:49 codesine sicp i think you'll have to learn with scheme

11:02:08 djh I'm going through SICP and wouldn't want to try it with CL instead of scheme

11:02:15 codesine ya

11:02:22 John[Lisbeth] try what?

11:02:23 codesine there's an interesting book

11:02:25 codesine the little schemer

11:02:34 ecraven is there that much difference? apart from continuations and TCO?

11:02:37 codesine it was first the little lisper, i think it has the translations of the lisp code in it as well

11:02:46 ecraven lisp-1 vs. lisp-2 isn't that hard to work around

11:03:00 malice ecraven: funcall makes code look more ugly imho

11:04:02 codesine I haven't looked at sicp, I did that practical common lisp, random reading of source code, but am thinking about reading the rest of the series to the little schemer

11:04:22 ecraven malice: indeed, but it isn't too bad, most things would work fine

11:08:33 Cymew There are multiple attempts to work through SICP with CL, and people have blogged about it. Might be worth searching for.

11:08:56 codesine sicp is recommended for thinking in lisp right?

11:09:19 Cymew It's a good book about Computer Science.

11:09:23 codesine oh

11:10:13 djh I'm finding it remarkably informative about many things, including Lisp

11:11:15 djh tho I think so far the books I've had the most "aha!" moments with have been PCL and AMOP

11:11:27 djh lisp-related moments, that is

11:11:33 codesine hey djh, did you read the little schemer / seasoned schemer / reasoned schemer

11:11:57 codesine it has scheme in the name, but all the snippets have translations for lisp (as it was originally for lisp (i think?))

11:11:59 djh I read the little lisper, at some point I must get around to the little schemer

11:12:16 codesine I just started the little lisper + almost done with PCL

11:12:41 codesine I'll prob look at a little of sicp once I have a clear head again (:

11:13:31 djh I largely put learning CL on hold to work through SICP, looking forward to coming back to it later :)

11:13:41 codesine I see

11:14:21 malice djh: AMOP?

11:14:25 djh reading AMOP at the moment on my commute tho, to keep my hand in :)

11:14:34 djh the Art of the Metaobject Protocol

11:15:34 djh been meaning to read up on MOPs since I started using Perl's, finally found the time courtesy of long train rides :)

11:20:48 malice I guess I've never used MOP; any short description why is it useful?

11:21:37 djh TBH I'm reading it because it's interesting rather than out of any specific use case

11:22:56 djh The general use-case is "you want OO to work differently than the default" such as different inheritance rules, or different ways of storing data in slots

11:26:44 pillton The design pattern MOP uses is very good to know. It works very well.

11:28:16 pillton Common Lisp HyperSpec: The very definition of class.

11:28:29 pillton Is there a story behind "The very definition of class"?

11:29:26 Shinmera It's a pun

11:30:08 lieven CL was the first ANSI standardized OO language

11:31:38 djh It kind of sums up the CL attitude for me, that when they needed to work out what OO rules to go with, they just said "let's invent a way to support *all* of them!" :)

11:44:25 codesine generally speaking, is it preferable to use (list 1 2 3) vs '(1 2 3)? Is it considered more idiomatic by the community?

11:44:58 codesine or just a matter of preference?

11:45:50 Shinmera If you modify literals, bad juju happens

11:46:12 Shinmera But if you know you won't modify it, it's fine.

11:49:46 malice codesine: Moreover, you can do (list 1 (evenp 2) 3), but '(1 (evenp 2) 3) won't work(in the same way as #'list one); you'd need `(1 ,(evenp 2) 3), or list.

11:50:01 |3b| LIST makes a new list every time, while '() might be combined with other similar lists during compilation (one reason you shouldn't modify it)

11:50:39 |3b| so in addition to the problem of modification, they behave differently in regards to identity as well

12:05:45 John[Lisbeth] I am reading the gigamonkeys and some of the emacs keybinds from slime are not working

12:06:07 John[Lisbeth] for example it says that c-c c-q should invoke slime-close-parens-at-point

12:06:21 John[Lisbeth] But my setup says that is unbound

12:06:47 |3b| some things have changed since the book was written :/

12:07:16 |3b| (though for that specific thing, you might look at paredit or similar, that close parens when you open them, so you always have both)

12:11:45 John[Lisbeth] There's another one I can't do called slime-compile-defun

12:11:49 John[Lisbeth] is that one important?

12:13:22 ecraven John[Lisbeth]: try C-h w (M-x where-is)

12:13:23 |3b| that one should work, at least in lisp files (the bindings are different in repl)

12:13:30 ecraven that tells you to what a command is bound

12:13:39 ecraven to

12:14:28 |3b| * uses C-c C-c a lot (which is slime-compile-defun in lisp files if slime is working)

14:11:21 drmeister Why in emacs does hitting "g" in the *compilation* window restart the compilation (proper behavior) and other times hitting "g" inserts a "g".

14:15:51 igam drmeister: in some modes, we use direct keys for a smoother user interface. In those buffers, it would be meaningless to insert character or to edit in general. So those keys are free for other commands.

14:16:39 Shinmera I think he means that in the *compilation* window it sometimes inserts a g and sometimes does not

14:17:06 drmeister igam: Right, but the *compilation* windows sometimes get into a mode where "g" inserts a "g"

14:17:08 drmeister Right

14:17:16 drmeister igam: Thanks for responding by the way.

14:17:16 igam drmeister: thijs would be a bug.

14:17:19 Shinmera Is the mode line different when that happens?

14:17:46 jackdaniel igam: (or a feature ;-)

14:17:51 igam You can always change the mode to an editing mode. But it would not be the normal workflow.

14:18:08 drmeister Shinmera: Thanks - good question - I can't tell right now hang on.

14:18:34 igam Notice that I would want to be able to input into *compilation* buffer, since sometimes makefiles are interactive. But it's not possible, and one has to run those make in shell instead.

14:18:52 drmeister Dang it! It just switched back to the proper behavior.

14:20:17 drmeister It happens a couple of times a day - I'll make a note to check next time.

14:21:06 drmeister Whew - all of the compile errors are now out after making closures first class objects.

14:21:21 drmeister Now comes the fun part of debugging it.

14:21:35 igam drmeister: when this happens use M-x view-lossage RET = C-h l ; it will show you what you just did, and will probably explain the behavior.

14:22:53 drmeister igam: Thanks!

14:28:32 i-love-iisp Is there a way to define a symbol besides nil, that would also evaluate to nil if used on a test? Like (if :my-nil 3 4) returning 4

14:28:32 drmeister Nice - the REPL started up first time after making lots of changes over the last couple of days.

14:29:30 drmeister i-love-iisp: What is the motivation behind that?

14:30:29 |3b| (defconstant +my-nil+ nil)?

14:30:56 |3b| or similarly with defparameter, if you don't care if someone can redefine it to something else

14:31:21 i-love-iisp I extended cl-json to convert the json null to nil, and the json false to the symbol :false

14:31:52 |3b| symbols in the keyword package are defined to evaluate to themselves, so ::my-nil evaluating to nil would be a bit odd, but you could probably manage it on most implementations with a bit of effort

14:32:14 drmeister Well, don't use a keyword - they evaluate to themselves

14:32:19 i-love-iisp but then i always have to do (if (eq my-var t) ...) instead of (if my-var), since :false evaluates to true

14:32:19 |3b| :my-nil i mean

14:32:36 drmeister So +false+

14:34:06 Shinmera You'll have to implement your own JS-like IF if you want that kind of ambiguity.

14:34:52 Shinmera Personally I don't like "weird shit" like that, but whatever.

14:34:55 i-love-iisp Hm, but yeah, i think what |3b| is saying should work. use a constant defined as nil instead of a keyword

14:35:05 Shinmera Then you'll lose distinction though

14:35:22 Shinmera null will be translated exactly to the same thing as false.

14:35:41 Shinmera So you might as well just use NIL for both.

14:35:44 i-love-iisp Yeah, i really need to work with json on this project though, and go back and forth, so i somehow need to represent [], false and null in lisp

14:36:08 i-love-iisp oh, ofcourse

14:36:26 i-love-iisp yes, it still doesn't work

14:36:30 |3b| note that a variable containing +my-nil+ won't evaluate to false though

14:36:31 Zhivago json:false, json:null, json:array.

14:37:31 |3b| you could implement a json-if that uses JS style logic

14:37:44 Zhivago (defconstant json:false 'json:false) ;)

14:38:26 Zhivago Then (json:if ...) and off you go.

14:39:13 i-love-iisp yes, i guess implementing an if that tests for those keywords works. I was thinking maybe there was a way I didn't know with defining a type subclassing nil or something

14:41:24 Zhivago CL isn't particularly extensible.

14:41:48 Zhivago But it does make it reasonably easy to write little dialects on top of itself.

14:47:38 vydd Shinmera: hey, can you point me to a project (yours or not) that you think is very well documented. what I'm mainly interested in is to see how people go about commentating on types and how people talk about CLOS related stuff, also, macros - the other things I think I can take from my experience with other languages

14:48:44 Shinmera vydd: ediware is generally well documented.

14:49:22 vydd Shinmera: ok, thanks :)

14:49:32 |3b| also CL itself

14:49:45 |3b| much of it is implemented as macros, and well documented

14:50:17 mood I like to think Birch, my IRC library, is pretty well documented in the README: https://github.com/jorams/birch

14:50:24 Shinmera vydd: Of my projects, I guess Qtools and North have the most docs.

14:50:48 vydd |3b|: ha. ok. as silly as it sounds, looking at those docs for examples didn't occur to me. thanks!

14:51:02 |3b| i guess the pretty printer does some CLOS stuff as well. Probably the most important part there that differs from other languages is documenting the protocols defined by the various generic functions

14:52:59 vydd mood: looks good (also, a very nice project name), thanks

14:53:51 Shinmera I should return to Colleen3 development myself, actually.

14:54:02 Shinmera But I'm kinda bummed out on it

15:18:29 John[Lisbeth] http://sprunge.us/TgLM

15:19:15 John[Lisbeth] It complains hello-world is not a valid name

15:19:30 |3b| no, it probably complains (hello-world) is not a valid name

15:19:38 sjl I'm reading PAIP and in one of these functions uses .,(some form that returns a list) inside a quasiquote

15:19:38 |3b| that's rather oddly formatted too

15:19:48 sjl is ., not essentially the same as ,@?

15:20:08 |3b| similar effect if you don't mutate any of the lists involved

15:20:16 |3b| and only works at end of list

15:20:30 |3b| ., is just normal consing . and , from `

15:20:37 John[Lisbeth] Why can't I put hello-world into parentheses?

15:20:39 sjl yeah

15:20:44 DeadTrickster am I being kicked out?

15:20:48 Bike it's like ,@ but uses nconc instead of append.

15:20:49 sjl so I'm wondering why he uses it if it's the same as ,@

15:20:50 |3b| John[Lisbeth]: because that's how the language is defined

15:20:57 DeadTrickster each time I'm clicking on his link channel tab closes

15:21:03 |3b| sjl: memory was more expensive long ago

15:21:07 DeadTrickster or it looks so

15:21:20 mood DeadTrickster: From here it just looks like you're leaving

15:21:22 sjl |3b|: yeah... he uses ,@ in a bunch of other places though

15:21:23 John[Lisbeth] |3b|: is there a dialect of lisp where I could write (((hello-world))), or an arbitrary number of parentheses?

15:21:34 John[Lisbeth] just out of curiosity

15:21:39 Bike why on earth would you want to do that

15:21:54 DeadTrickster parenception

15:21:56 John[Lisbeth] I like put parentheses around anything I can get away with

15:22:10 |3b| John[Lisbeth]: none that i know of, () mean specific things in lisps

15:22:10 Bike «is there a dialect of C where I can write puts((("hello world")))»

15:22:16 sjl John[Lisbeth]: (((hello-world))) is like saying hello-world()()() in a C-like language

15:22:34 sjl kind of

15:22:40 John[Lisbeth] I've never seen a syntax like that in a c language

15:22:43 |3b| similarly (t) probably won't work

15:22:52 John[Lisbeth] That doesn't make sense to me

15:22:56 |3b| it would be like 1() in C

15:23:11 John[Lisbeth] so parens only work if you have more than one argument?

15:23:11 |3b| T isn't a function in CL, just like 1 isn't a function in C, so you can't call it as one

15:23:24 sjl in C-like languages parens are used for both grouping (precedence-wise) and function calls

15:23:33 |3b| parens don't "work"

15:23:45 |3b| parens always make a list, regardless of how many you put

15:23:52 |3b| the part that doesn't work is /evaluating/ that list

15:24:33 |3b| CL specified how to evaluate a form that is a list, and those rules require that the first element either be a symbol naming an operator, or a list which is a valid lambda form

15:25:33 Cymew If you "like put parentheses around anything I can get away with" I wonder if you are trying to understand how lisp work, or if you are just fooling around to see what happens.

15:25:54 DeadTrickster >> parens always make a list

15:25:57 DeadTrickster not always

15:26:02 Cymew Both are valid endeavors, but one if more on topic here...

15:26:12 DeadTrickster but at least when default readtable is active

15:26:15 Cymew oops s/if/is/

15:26:34 |3b| DeadTrickster: true, i simplified a bit... they also don't make a list inside strings, when quoted, etc

15:26:47 DeadTrickster lets confuse him futher

15:28:59 |3b| John[Lisbeth]: CL divides things up into separate stages, first it READs the text of the program to create lisp objects like symbols, lists, etc, then it evaluates (or compiles) that lisp data, so it has to create a list for every () it sees since the READer doesn't know about semantic meaning of what it is reading

15:29:54 |3b| actually, i guess technically you could write (((hello))) in scheme, if (hello) returns a function of 0 arguments that returns a function of 0 arguments

15:30:09 DeadTrickster the same is true for cl

15:30:31 |3b| no, the only list form valid as an operator in CL is (lambda (...) ...)

15:30:54 DeadTrickster that's what I meant

15:30:58 |3b| John[Lisbeth]: but you couldn't wrap that in another () unless that last function also returned a function of 0 arguments (and so on)

15:31:00 DeadTrickster ((lambda () (print "qwe")))

15:31:11 ggole If you want useless parens, write a macro (useless-parens (((foo)))) that expands to (foo)

15:31:19 |3b| right, but you can't (((lambda ...)))

15:31:47 DeadTrickster we can't go deeper

15:32:13 DeadTrickster which is looking like unnecessary restriction actually to me

15:35:39 |3b| DeadTrickster: yeah, they could have defined arbitrary other behaviors for lists as first element of lists, but they didn't :)

15:35:51 |3b| sort of like how the selection of special operators is sort of arbitrary

15:36:23 DeadTrickster just evaluate first list element if it is function call it

15:36:26 DeadTrickster sounds simple

15:36:33 ggole Evaluating a list form but not a symbol would have been pretty whack

15:36:45 |3b| though then you get into the question if "if lists in operator position are evaluated normally, why aren't symbols?"

15:36:50 ggole And if you evaluate all positions, then you have a lisp-1

15:37:00 |3b| ^

15:37:10 ggole (Which is a perfectly fine design if you ask me, but is not the design of CL.)

15:37:43 |3b| right, and you can just as well type (funcall (...)) as ((...))

15:38:08 DeadTrickster yep

15:40:05 |3b| ((lambda ())) is probably mostly intended for macro/reader macro use

15:40:25 Shinmera (setf (fdefinition '​) #'funcall)( ​((lambda () (lambda () "hi")))) :^)

15:40:35 Shinmera ech, paste failed

15:40:42 jackdaniel ((lambda (hue) hue) :hue)

15:40:49 Shinmera (setf (fdefinition '​) #'funcall) (​((lambda () (lambda () "hi"))))

15:41:30 Shinmera Anyway, unicode fun.

15:41:55 Shinmera Surprise your friends, confound your enemies.

15:41:58 |3b| ah, 0width space... just shows up as _ here :p

15:42:38 |3b| * switched to a more limited font now that i'm back on a low-dpi monitor :(

15:43:08 jackdaniel I was wondering, how much of heresy would be evaluating the argument on first position only if it isn't a function name?

15:43:33 Shinmera * likes Consolas

15:43:51 Shinmera jackdaniel: That sounds really bad

15:43:55 |3b| * seems to recall thinking of some place where it wasn't obvious how to interpret that, but can't think of it at the moment

15:44:25 jackdaniel Shinmera: I think it depends on the point of view, you get lisp-1 fallback for nothing

15:44:28 John[Lisbeth] So you are saying it doesn't work in common lisp because it has multi-value namespace?

15:44:39 Shinmera jackdaniel: What if you have a local variable binding that is not a function?

15:45:04 |3b| Shinmera: bit ugly at 8pt, too big otherwise :/

15:45:05 jackdaniel Shinmera: like (let ((a #'car)) (a '(1 2 3))) ;?

15:45:34 |3b| much more options at 150+dpi than at <100 :(

15:45:41 Shinmera jackdaniel: No, like (defun foo ()) (let ((foo 0)) (foo)) => ??

15:46:03 jackdaniel Shinmera: *only* if there is no function in the functions namespace

15:46:11 jackdaniel I wrote that ,þ

15:46:15 |3b| John[Lisbeth]: it won't work because it is defined to not work, and even if it did work, probably wouldn't do what you seemed to want

15:46:30 Shinmera jackdaniel: Then you're forcing the programmer to be aware of the full function namespace at all times. Pretty awful.

15:47:09 John[Lisbeth] If it makes sense I basically want everything inbetween the space character to be surrounded by parentheses unless they are between quotation marks.

15:47:28 John[Lisbeth] I am not sure why I feel it should work this way

15:47:34 |3b| John[Lisbeth]: possibly CL isn't the language for you then :(

15:47:50 John[Lisbeth] I like my syntax to be very regular is all

15:48:01 jackdaniel Shinmera: I would say, that using a symbol in the first position would be a bad practice. Either way, just an idea

15:48:10 jackdaniel I don't need it that bad to implement it

15:48:10 |3b| CL is regular, if you understand the evaluation rules

15:48:25 John[Lisbeth] the perl people swear perl is regular

15:48:39 |3b| but at the semantic level rather than the textual level

15:49:01 |3b| so if the textual result doesn't work for you, you might want something else

15:49:40 |3b| * doesn't like python for the way it uses whitespace, so i can't complain if you don't like the way CL uses () :)

15:49:53 John[Lisbeth] I think it is not totally necessary for me to use parentheses that way.

15:50:24 jackdaniel there was some project messing with the readtable to reduce number of parens in favour of spaces and some inference

15:50:31 jackdaniel don't remember the name though

15:50:55 |3b| possibly you could define some reader macros to make CL surface syntax more palatable, but at that point you aren't really using CL, and no CL programmer would want to read it :)

15:51:15 |3b| jackdaniel: i think there have been a few

15:51:30 |3b| some repls that let you drop the outer set of parens too

15:57:38 jackdaniel one problem with evaluating symbol which isn't in the function namespace would be (let ((a 'b) (b 'a)) (a 3))

16:00:36 ggole But neither of those is in the function namespace?

16:01:21 |3b| seems you would just error after the first didn't evaluate to a function

16:01:48 jackdaniel |3b|: depens how would you implement the evaluator

16:02:42 |3b| yeah, but doesn't seem like a particularly useful way to implement it

16:02:59 fiddlerwoaroof jackdaniel: I think you're talking about readable?

16:03:09 fiddlerwoaroof http://readable.sourceforge.net/

16:03:32 jackdaniel (defun evaluate (...) ... (if (functionp (car sexp) (funcall (car sexp))) (funcall (evaluate (car sexp))) ...) would burp in the infinite recursion

16:03:40 |3b| that way of interpreting it sounds messy, what would (let ((a 'let)) (a 1)) do?

16:04:00 |3b| or a macro or whatever

16:04:29 Ulster ** NICK fkac

16:04:31 |3b| you hit something with a definition in the function namespace, but that isn't a function

16:04:58 |3b| or worse (let ((a 'let) (let 'list)) (a 1))

16:04:59 jackdaniel * doesn't think, it's revelant to this particular problem

16:05:05 jackdaniel fiddlerwoaroof: yeah

16:05:30 fiddlerwoaroof I've thought that would be useful to use as an extension language for a lisp program

16:05:52 fiddlerwoaroof Since the syntax looks more "normal" to most people than making them right lisp

16:05:54 |3b| (let 1) instead of (a 1) would see the special operator LET and error, so seems odd the recursive expansion version would behave differently

16:05:59 jackdaniel |3b|: if symbol isn't found in the function namespace, then the first-position element is evaluated, and if it evaluates to the function, then it is funcalled

16:06:20 jackdaniel so fallback is indifferent from (funcall x …)

16:06:30 fiddlerwoaroof jackdaniel: couldn't that run into double-evaluation problems?

16:06:57 jackdaniel fiddlerwoaroof: I didn't think this through, maybe

16:06:58 |3b| right, but your example evaluated to a symbol, so in that case it would just error?

16:07:00 fiddlerwoaroof hmm, I guess it would be invalid syntax otherwise, so maybe not.

16:07:53 jackdaniel |3b|: given implementation something like (defun evaluate ...) I shown would fall into infinite recursion. It's a problem which might be avoided I think

16:07:58 |3b| * assumed you meant an evaluation rule where it would expand A to B, see there was no function named B, then expand that to A, and repeat indefinitely

16:08:26 fiddlerwoaroof It might be better to just macroexpand the first argument.

16:08:26 jackdaniel yes, that would be a result of clumsy implementation :)

16:08:42 |3b| so in my (a 'let) example, would it find LET in the function namespace, or ignore it because it doesn't name a function?

16:09:15 jackdaniel is let in the function namespace?

16:09:33 jackdaniel but no, it will only verify, if a is in the function namespace

16:09:34 |3b| it is usable as an operator

16:09:39 jackdaniel if not, it would evaluate it

16:09:53 jackdaniel and if it evaluates to function, then it calls it with the rest arguments

16:10:32 |3b| * thinks the discussion is a good enough reason for the specified behavior, too complicated to come up with all the details for a feature that wouldn't be used much

16:10:59 |3b| that part makes sense, but i'd assume you finish that with "and in any other case it errors"

16:11:56 |3b| not "then it tries to evaluate that again, but only if it is a variable that doesn't evaluate to itself"

16:12:35 |3b| since (:foo) would always evaluate to :foo rather than a function, similarly 1 would always evaluate to 1

16:12:39 fiddlerwoaroof * http://paste.lisp.org/+6O2Z

16:13:02 jackdaniel yeah, you are right, I just came up with the potential implementation problem

16:14:14 jackdaniel fiddlerwoaroof: nice :)

16:15:02 fiddlerwoaroof I was just thinking about how I kinda like the way Arc lets you treat data structures as functions.

16:15:11 jackdaniel (defmacro defht (name) `(progn (defparameter ,name (make-hash-table)) (defmacro ,name (key) `(gethash ,key ,,name)))

16:15:31 fiddlerwoaroof Yeah, that's the next step.

16:15:59 jackdaniel but given that the evaluator is extensible, you could just treat data separately

16:16:16 jackdaniel "extensible evaluator" sounds cool ^_^

16:17:15 jackdaniel marzenie scietej glowy, it would fail bad given that you may actually want to compile it (the extensible evaluator thing)

16:43:28 rudolfochrist

16:43:31 rudolfochrist [Wed Nov 18 2015]

16:43:34 rudolfochrist *** Topic for #lisp: Common Lisp, the #1=(programmable . #1#) programming

16:43:39 rudolfochrist language <http://cliki.net/> <http://paste.lisp.org/new>

16:43:42 rudolfochrist logs:<http://ccl.clozure.com/irc-logs/lisp/>|contact op if muted|SBCL

16:43:45 rudolfochrist 1.3.0, ASDF 3.1.6, CMUCL 21a, ECL 16.0.0, CFFI 0.16.0, cl-launch 4.1.4,

16:43:46 Shinmera Thanks for that.

16:43:48 rudolfochrist flexi-streams 1.0.15, Hunchentoot 1.2.34, Drakma 2.0.1 [21:22]

16:43:48 rudolfochrist *** #lisp: topic set by fe[nl]ix!~quassel@pdpc/supporter/professional/fenlix,

16:43:51 rudolfochrist 2015-11-02 19:31:05

16:43:54 rudolfochrist *** Users on #lisp: rudolfochrist nikki93 moei jinxter Bicyclidine arescorpio

16:43:58 rudolfochrist grouzen pt1 briantrice sjl jaykru mitzelflick karswell ecraven bb010g

16:44:01 rudolfochrist armour EvW pyon hiroakip cadadar dougk_ fourier Mon_Ouie cyraxjoe atgreen

16:44:04 rudolfochrist gravicappa scymtym_ badkins anti-freeze attila_lendvai lnostdal abaugher

16:44:08 rudolfochrist DeadTrickster_ Draz fantomik preacherAKAnd zacharias vlnx Guest30769

16:44:12 rudolfochrist zygentoma TMM dbrock- duggiefresh oem Lord_Nightmare m0li aphprentice

16:44:15 rudolfochrist defaultxr shka_ _sjs OrangeShark knobo

16:44:18 rudolfochrist *** Users on #lisp: varjagg newdan Yanez psy_ Jesin mordocai et8 antonv

16:44:18 rudolfochrist shookees LiamH lancetw blubjr` gilez oleo Guest32265 linoge warweasle

16:44:21 rudolfochrist knobo_ Oladon lxpz ogamita vydd kobain sdothum Lord_of_Life quazimodo

16:44:22 sjl oh ffs

16:44:24 rudolfochrist Beluki papachan larme sz0 kjeldahl dlowe wemeetagain Nikotiini edgar-rft

16:44:28 rudolfochrist znpy araujo adhoc Moyst eazar001 dkcl gargaml gingerale ebrasca stepnem

16:44:31 rudolfochrist mishoo_ Yuuhi Shinmera pjb UtkarshRay srcerer nowhere_man Whymind zaquest

16:44:34 rudolfochrist Meow-J lemonpepper24 sekrit Zotan Bike aap Xach

16:44:36 ChanServ ** MODE +o fe[nl]ix

16:45:34 zygentoma \o/

16:45:44 oleo o/

16:47:09 |3b| where do people still find that version of erc or whatever it is that does that?

16:47:25 jackdaniel Nov 2015? hrm :)

16:47:53 |3b| hmm, that does sound a bit different

16:51:04 sjl eight megs and constantly spamming

16:51:05 |3b| that person apparently has had the same problem before with same chunk of log :p

16:51:40 Bicyclidine there is the theory of the möbius, a region of chat in which time becomes a loop

17:03:51 sweater ** NICK Guest95255

17:13:37 beach Good evening everyone!

17:14:03 oleo evening beach

17:14:33 Shinmera Hello beach

17:16:58 beach It would be great if someone would take on the maintenance of CL-VECTORS. Of course, I haven't (yet) asked whether the author would be interested. In particular, the documentation needs to be brought up to date, and missing things needs to be added to it.

17:18:43 warweasle beach: Sorry, I'm using cairo and freeimage.

17:21:03 rudolfochrist

17:21:03 rudolfochrist [Wed Nov 18 2015]

17:21:03 rudolfochrist *** Topic for #lisp: Common Lisp, the #1=(programmable . #1#) programming

17:21:03 rudolfochrist language <http://cliki.net/> <http://paste.lisp.org/new>

17:21:05 rudolfochrist logs:<http://ccl.clozure.com/irc-logs/lisp/>|contact op if muted|SBCL

17:21:08 rudolfochrist 1.3.0, ASDF 3.1.6, CMUCL 21a, ECL 16.0.0, CFFI 0.16.0, cl-launch 4.1.4,

17:21:11 rudolfochrist flexi-streams 1.0.15, Hunchentoot 1.2.34, Drakma 2.0.1 [21:22]

17:21:14 rudolfochrist *** #lisp: topic set by fe[nl]ix!~quassel@pdpc/supporter/professional/fenlix,

17:21:18 rudolfochrist 2015-11-02 19:31:05

17:21:21 rudolfochrist *** Users on #lisp: rudolfochrist nikki93 moei jinxter Bicyclidine arescorpio

17:21:25 rudolfochrist grouzen pt1 briantrice sjl jaykru mitzelflick karswell ecraven bb010g

17:21:28 rudolfochrist armour EvW pyon hiroakip cadadar dougk_ fourier Mon_Ouie cyraxjoe atgreen

17:21:29 rudolfochrist gravicappa scymtym_ badkins anti-freeze attila_lendvai lnostdal abaugher

17:21:48 beach warweasle: I understand. When I said "someone", I meant "someone using Common Lisp".

17:21:56 fe[nl]ix so 'twasn't a mistake

17:22:20 edgar-rft using Common Lisp never is a mistake

17:22:30 moore33 beach: He means that he calls those libraries from Common Lisp.

17:22:38 beach edgar-rft: He is talking about rudolfochrist.

17:22:50 fe[nl]ix because those not using CL are otherones, not someones

17:22:53 beach moore33: I know perfectly well what he means.

17:23:52 fe[nl]ix beach: you can ask the author to transfer ownership to sharplispers

17:24:22 beach fe[nl]ix: Good point. I'll start by asking him whether he would be interested in maintaining it himself.

17:24:34 moore33 beach: Heh, well I understand that to mean "using Common Lisp," though perhaps not using it for rasterization.

17:25:06 beach Yes, I am fully aware of the current situation.

17:25:32 fe[nl]ix we can of course import the repository to github/sharplispers but I think Xach wishes the author's permission to switch the Quicklisp source to that new repo

17:25:55 beach fe[nl]ix: I'll try to deal with the issue. Thanks for the suggestion.

17:29:22 beach moore33: Sorry for sounding snooty. I am in a rotten mood.

17:30:51 moore33 beach: Oh! Sorry to hear that.

17:31:46 beach moore33: Thanks!

17:38:15 myrkraverk Is it possible, to detect with lisp code, if it's running through a hashbang? As in #!/usr/bin/sbcl --script

17:41:52 fiddlerwoaroof myrkraverk: you might check *features*

17:41:59 fiddlerwoaroof But, it'll be implementation dependent.

17:42:49 myrkraverk ok

17:43:38 fiddlerwoaroof Anyways, I don't see anything in *features*

17:46:28 xrosnight ** NICK hoodex

17:46:28 myrkraverk *nod*